Review of survival analyses published in cancer journals

IELSG38: phase II trial of front-line chlorambucil plus subcutaneous rituximab induction and maintenance in mucosa-associated lymphoid tissue lymphoma

The IELSG38 trial was conducted to investigate the effects of subcutaneous (SC) rituximab on the complete remission (CR) rate and the benefits of SC rituximab maintenance in patients with extranodal marginal zone lymphoma (MZL) who received front-line treatment with chlorambucil plus rituximab. Study treatment was an induction phase with oral chlorambucil 6 mg/m2/day on weeks 1-6, 9-10, 13-14, 17-18, and 21-22, and intravenous rituximab 375 mg/m2 on day 1 of weeks 1-4, and 1,400 mg SC on weeks 9, 13, 17, and 21. Then, a maintenance phase followed with rituximab administered at 1,400 mg SC every two months for two years. Of the 112 patients enrolled, 109 were evaluated for efficacy. The CR rates increased from 52% at the end of the induction phase to 70% upon completion of the maintenance phase. With a median follow-up of 5.8 years, the 5-year event-free, progression-free, and overall survival rates were 87% (95% CI: 78-92), 84% (95% CI: 75-89), and 93% (95% CI: 86-96), respectively. The most common grade ≥3 toxicities were neutropenia (33%) and lymphocytopenia (16%). Six patients experienced treatment-related serious adverse events, including fever of unknown origin, sepsis, pneumonia, respiratory failure, severe cerebellar ataxia, and fatal acute myeloid leukemia. The trial showed that SC rituximab did not improve the CR rate at the conclusion of the induction phase, which was the main endpoint. Nevertheless, SC rituximab maintenance might have facilitated long-term disease control, potentially contributing to enhanced event-free and progression-free survival.

The use and abuse of survival analysis and Kaplan-Meier curves in surgical trials

BACKGROUND

Survival analysis based on Cox regression and Kaplan-Meier curves, initially devised for oncology trials, have frequently been used in other contexts where fundamental statistical assumptions (such as a constant hazard ratio) are not satisfied. This is almost always the case in trials that compare surgery with medical management.

METHODS

We review a trial that compared extracranial-intracranial bypass surgery (EC-IC bypass) with medical management (MM) of patients with symptomatic occlusion of the carotid or middle cerebral artery, where it was claimed that surgery was of no benefit. We discuss a hypothetical study and review other neurovascular trials which have also used survival analysis to compare results.

RESULTS

The trial comparing EC-IC bypass and MM did not satisfy the fundamental proportional hazard assumption necessary for valid analyses. This was also the case for two prior EC-IC bypass trials, as well as for other landmark neurovascular studies, such as the trials comparing endarterectomy with MM for carotid stenoses, or for the trial that compared intervention and MM for unruptured brain arteriovenous malformations. While minor deviations may have little effect on large trials, it may be impossible to show the benefits of surgery when trial size is small and deviations large.

CONCLUSION

Survival analyses are inappropriate in RCTs comparing surgery with conservative management, unless survival is calculated after the postoperative period. Alternative ways to compare final clinical outcomes, using for example a fixed follow-up period, should be planned for preventive surgical trials that compare intervention with conservative management.

Lipoprotein Insulin Resistance Score and Mortality Risk Stratification in Heart Failure

BACKGROUND

Higher total serum cholesterol is associated with lower mortality in heart failure. Evaluating associations between lipoprotein subfractions and mortality among people with heart failure may provide insights into this observation.

METHODS

We prospectively enrolled a community cohort of people with heart failure from 2003 to 2012 and assessed vital status through 2021. Plasma collected at enrollment was used to measure lipoprotein subfractions via nuclear magnetic resonance spectroscopy. A composite score of 6 lipoprotein subfractions was generated using the lipoprotein insulin resistance index (LP-IR) algorithm. Using covariate-adjusted proportional hazards regression models, we evaluated associations between LP-IR score and all-cause mortality.

RESULTS

Among 1382 patients with heart failure (median follow-up 13.9 years), a one-standard-deviation (SD) increment in LP-IR score was associated with lower mortality (hazard ratio [HR] 0.93; 95% confidence interval [CI], 0.97-0.99). Among LP-IR parameters, mean high-density lipoprotein (HDL) particle size was significantly associated with lower mortality (HR per 1-SD decrement in mean HDL particle size = 0.83; 95% CI, 0.78-0.89), suggesting that the inverse association between LP-IR score and mortality may be driven by smaller mean HDL particle size.

CONCLUSIONS

LP-IR score was inversely associated with mortality among patients with heart failure and may be driven by smaller HDL particle size.

Postchemotherapy Retroperitoneal Lumpectomy in Patients with Nonseminoma Testicular Cancer: A Nationwide Study

BACKGROUND AND OBJECTIVE

Optimal treatment outcomes in patients with metastatic nonseminoma testicular cancer are achieved with chemotherapy and subsequent surgery in cases with residual tumor. In Denmark, postchemotherapy retroperitoneal lumpectomy (RPLP) is performed in patients with residual tumors >1 cm. There is a need to clarify whether this surgical method provides acceptable treatment results. Our objective was to describe morbidity and oncological outcomes of postchemotherapy RPLP.

METHODS

This was a retrospective population-based multicenter study including patients with nonseminoma testicular cancer and postchemotherapy RPLP performed in Denmark between 1990 and 2015. A total of 219 patients were eligible, with median follow-up of 19 yr. Postoperative complications were evaluated according to the Clavien-Dindo classification. The cumulative incidence of recurrence inside or outside the borders of a bilateral surgical template, progression-free survival (PFS), and overall survival estimates were calculated using the Kaplan-Meier method.

KEY FINDINGS AND LIMITATIONS

After median follow-up of 19 yr, 31/219 patients (14%) experienced a surgical complication, of which 5% were Clavien-Dindo grade ≥III. In total, 37 patients experienced a recurrence. The 5-yr, 10-yr, and 20-yr cumulative risk of recurrence inside a bilateral template was 4.3%, 5.9%, and 5.9%, respectively. The 10-yr PFS rate was 83% and the 10-yr overall survival rate was 96%. The main limitation of the study is the retrospective design.

CONCLUSIONS AND CLINICAL IMPLICATIONS

With few patients experiencing a major postoperative complication and a 10-yr cumulative rate of 5.9% for recurrence inside a bilateral surgical template, postchemotherapy RPLP appears to be a safe alternative to template surgery for disseminated nonseminoma.

PATIENT SUMMARY

We looked at minimal surgery to remove tumor tissue remaining after chemotherapy in patients with testicular cancer. We found a low frequency of complications, tumor recurrence, and death.

ECG-surv: A deep learning-based model to predict time to 1-year mortality from 12-lead electrocardiogram

BACKGROUND

Electrocardiogram (ECG) abnormalities have demonstrated potential as prognostic indicators of patient survival. However, the traditional statistical approach is constrained by structured data input, limiting its ability to fully leverage the predictive value of ECG data in prognostic modeling.

METHODS

This study aims to introduce and evaluate a deep-learning model to simultaneously handle censored data and unstructured ECG data for survival analysis. We herein introduce a novel deep neural network called ECG-surv, which includes a feature extraction neural network and a time-to-event analysis neural network. The proposed model is specifically designed to predict the time to 1-year mortality by extracting and analyzing unique features from 12-lead ECG data. ECG-surv was evaluated using both an independent test set and an external set, which were collected using different ECG devices.

RESULTS

The performance of ECG-surv surpassed that of the Cox proportional model, which included demographics and ECG waveform parameters, in predicting 1-year all-cause mortality, with a significantly higher concordance index (C-index) in ECG-surv than in the Cox model using both the independent test set (0.860 [95% CI: 0.859- 0.861] vs. 0.796 [95% CI: 0.791- 0.800]) and the external test set (0.813 [95% CI: 0.807- 0.814] vs. 0.764 [95% CI: 0.755- 0.770]). ECG-surv also demonstrated exceptional predictive ability for cardiovascular death (C-index of 0.891 [95% CI: 0.890- 0.893]), outperforming the Framingham risk Cox model (C-index of 0.734 [95% CI: 0.715-0.752]).

CONCLUSION

ECG-surv effectively utilized unstructured ECG data in a survival analysis. It outperformed traditional statistical approaches in predicting 1-year all-cause mortality and cardiovascular death, which makes it a valuable tool for predicting patient survival.

Frailty and Metabolic Vulnerability in Heart Failure: A Community Cohort Study

BACKGROUND

Frailty is common in heart failure (HF) and is associated with death but not routinely captured clinically. Frailty is linked with inflammation and malnutrition, which can be assessed by a novel plasma multimarker score: the metabolic vulnerability index (MVX). We sought to evaluate the associations between frailty and MVX and their prognostic impact.

METHODS AND RESULTS

In an HF community cohort (2003-2012), we measured frailty as a proportion of deficits present out of 32 physical limitations and comorbidities, MVX by nuclear magnetic resonance spectroscopy, and collected extensive longitudinal clinical data. Patients were categorized by frailty score (≤0.14, >0.14 and ≤0.27, >0.27) and MVX score (≤50, >50 and ≤60, >60 and ≤70, >70). Cox models estimated associations of frailty and MVX with death, adjusted for Meta-Analysis Global Group in Chronic Heart Failure (MAGGIC) score and NT-proBNP (N-terminal pro-B-type natriuretic peptide). Uno's C-statistic measured the incremental value of MVX beyond frailty and clinical factors. Weibull's accelerated failure time regression assessed whether MVX mediated the association between frailty and death. We studied 985 patients (median age, 77; 48% women). Frailty and MVX were weakly correlated (Spearman's ρ=0.21). The highest frailty group experienced an increased rate of death, independent of MVX, MAGGIC score, and NT-proBNP (hazard ratio, 3.3 [95% CI, 2.5-4.2]). Frailty improved Uno's c-statistic beyond MAGGIC score and NT-proBNP (0.69-0.73). MVX only mediated 3.3% and 4.5% of the association between high and medium frailty groups and death, respectively.

CONCLUSIONS

In this HF cohort, frailty and MVX are weakly correlated. Both independently contribute to stratifying the risk of death, suggesting that they capture distinct domains of vulnerability in HF.

Prognostic impact of HER2 biomarker levels in trastuzumab-treated early HER2-positive breast cancer

BACKGROUND

Overexpression of human epidermal growth factor receptor 2 (HER2) caused by HER2 gene amplification is a driver in breast cancer tumorigenesis. We aimed to investigate the prognostic significance of manual scoring and digital image analysis (DIA) algorithm assessment of HER2 copy numbers and HER2/CEP17 ratios, along with ERBB2 mRNA levels among early-stage HER2-positive breast cancer patients treated with trastuzumab.

METHODS

This retrospective study comprised 371 early HER2-positive breast cancer patients treated with adjuvant trastuzumab, with HER2 re-testing performed on whole tumor sections. Digitized tumor tissue slides were manually scored and assessed with uPath HER2 Dual ISH image analysis, breast algorithm. Targeted ERBB2 mRNA levels were assessed by the Xpert® Breast Cancer STRAT4 Assay. HER2 copy number and HER2/CEP17 ratio from in situ hybridization assessment, along with ERBB2 mRNA levels, were explored in relation to recurrence-free survival (RFS).

RESULTS

The analysis showed that patients with tumors with the highest and lowest manually counted HER2 copy number levels had worse RFS than those with intermediate levels (HR = 2.7, CI 1.4-5.3, p = 0.003 and HR = 2.1, CI 1.1-3.9, p = 0.03, respectively). A similar trend was observed for HER2/CEP17 ratio, and the DIA algorithm confirmed the results. Moreover, patients with tumors with the highest and the lowest values of ERBB2 mRNA had a significantly worse prognosis (HR = 2.7, CI 1.4-5.1, p = 0.003 and HR = 2.8, CI 1.4-5.5, p = 0.004, respectively) compared to those with intermediate levels.

CONCLUSIONS

Our findings suggest that the association between any of the three HER2 biomarkers and RFS was nonlinear. Patients with tumors with the highest levels of HER2 gene amplification or ERBB2 mRNA were associated with a worse prognosis than those with intermediate levels, which is of importance to investigate in future clinical trials studying HER2-targeted therapy.

The Metabolic Vulnerability Index: A Novel Marker for Mortality Prediction in Heart Failure

BACKGROUND

Inflammation and protein energy malnutrition are associated with heart failure (HF) mortality. The metabolic vulnerability index (MVX) is derived from markers of inflammation and malnutrition and measured by nuclear magnetic resonance spectroscopy. MVX has not been examined in HF.

OBJECTIVES

The authors sought to examine the prognostic value of MVX in patients with HF.

METHODS

The authors prospectively assembled a population-based cohort of patients with HF from 2003 to 2012 and measured MVX scores with a nuclear magnetic resonance scan from plasma collected at enrollment. Patients were divided into 4 MVX score groups and followed until March 31, 2021.

RESULTS

The authors studied 1,382 patients (median age: 78 years; 48% women). The median MVX score was 64.6. Patients with higher MVX were older, more likely to be male, have atrial fibrillation, have higher NYHA functional class, and have HF duration of >18 months. Higher MVX was associated with mortality independent of Meta-analysis Global Group in Chronic Heart Failure score, ejection fraction, and other prognostic biomarkers. Compared to those with the lowest MVX, the HRs for MVX groups 2, 3, and 4 were 1.2 (95% CI: 0.9-1.4), 1.6 (95% CI: 1.3-2.0), and 1.8 (95% CI: 1.4-2.2), respectively (Ptrend < 0.001). Measures of model improvement document the added value of MVX in HF for classifying the risk of death beyond the Meta-analysis Global Group in Chronic Heart Failure score and other biomarkers.

CONCLUSIONS

In this HF community cohort, MVX was strongly associated with mortality independently of established clinical factors and improved mortality risk classification beyond clinically validated markers. These data underscore the potential of MVX to stratify risk in HF.

Isolating the effect of confounding from the observed survival benefit of screening participants - a methodological approach illustrated by data from the German mammography screening programme

BACKGROUND

Mammography screening programmes (MSP) aim to reduce breast cancer mortality by shifting diagnoses to earlier stages. However, it is difficult to evaluate the effectiveness of current MSP because analyses can only rely on observational data, comparing women who participate in screening with women who do not. These comparisons are subject to several biases: one of the most important is self-selection into the MSP, which introduces confounding and is difficult to control for. Here, we propose an approach to quantify confounding based on breast cancer survival analyses using readily available routine data sources.

METHODS

Using data from the Cancer Registry of North Rhine-Westphalia, Germany, we estimate the relative contribution of confounding to the observed survival benefit of participants of the German MSP. This is accomplished by comparing non-participants, participants with screen-detected and participants with interval breast cancers for the endpoints "death from breast cancer" and "death from all causes other than breast cancer" - the latter being assumed to be unrelated to any MSP effect. By using different contrasts, we eliminate the effects of stage shift, lead and length time bias. The association of breast cancer detection mode with survival is analysed using Cox models in 68,230 women, aged 50-69 years, with breast cancer diagnosed in 2006-2014 and followed up until 2018.

RESULTS

The hazard of dying from breast cancer was lower in participants with screen-detected cancer than in non-participants (HR = 0.21, 95% CI: 0.20-0.22), but biased by lead and length time bias, and confounding. When comparing participants with interval cancers and non-participants, the survival advantage was considerably smaller (HR = 0.62, 95% CI: 0.58-0.66), due to the elimination of stage shift and lead time bias. Finally, considering only mortality from causes other than breast cancer in the latter comparison, length time bias was minimised, but a survival advantage was still present (HR = 0.63, 95% CI: 0.56-0.70), which we attribute to confounding.

CONCLUSIONS

This study shows that, in addition to stage shift, lead and length time bias, confounding is an essential component when comparing the survival of MSP participants and non-participants. We further show that the confounding effect can be quantified without explicit knowledge of potential confounders by using a negative control outcome.

Circulating ketone bodies and mortality in heart failure: a community cohort study

Background

The relationship between ketone bodies (KB) and mortality in patients with heart failure (HF) syndrome has not been well established.

Objectives

The aim of this study is to assess the distribution of KB in HF, identify clinical correlates, and examine the associations between plasma KB and all-cause mortality in a population-based HF cohort.

Methods

The plasma KB levels were measured by nuclear magnetic resonance spectroscopy. Multivariable linear regression was used to examine associations between clinical correlates and KB levels. Proportional hazard regression was employed to examine associations between KB (represented as both continuous and categorical variables) and mortality, with adjustment for several clinical covariates.

Results

Among the 1,382 HF patients with KB measurements, the median (IQR) age was 78 (68, 84) and 52% were men. The median (IQR) KB was found to be 180 (134, 308) μM. Higher KB levels were associated with advanced HF (NYHA class III-IV) and higher NT-proBNP levels (both P < 0.001). The median follow-up was 13.9 years, and the 5-year mortality rate was 51.8% [95% confidence interval (CI): 49.1%-54.4%]. The risk of death increased when KB levels were higher (HRhigh vs. low group 1.23; 95% CI: 1.05-1.44), independently of a validated clinical risk score. The association between higher KB and mortality differed by ejection fraction (EF) and was noticeably stronger among patients with preserved EF.

Conclusions

Most patients with HF exhibited KB levels that were consistent with those found in healthy adults. Elevated levels of KB were observed in patients with advanced HF. Higher KB levels were found to be associated with an increased risk of death, particularly in patients with preserved EF.

Explainable Artificial Intelligence to Identify Dosimetric Predictors of Toxicity in Patients with Locally Advanced Non-Small Cell Lung Cancer: A Secondary Analysis of RTOG 0617

PURPOSE

Dosimetric predictors of toxicity in patients treated with definitive chemoradiation for locally advanced non-small cell lung cancer are often identified through trial and error. This study used machine learning (ML) and explainable artificial intelligence to empirically characterize dosimetric predictors of toxicity in patients treated as part of a prospective clinical trial.

METHODS AND MATERIALS

A secondary analysis of the Radiation Therapy Oncology Group (RTOG) 0617 trial was performed. Multiple ML models were trained to predict grade ≥3 pulmonary, cardiac, and esophageal toxicities using clinical and dosimetric features. Model performance was evaluated using the area under the curve (AUC). The best performing model for each toxicity was explained using the Shapley Additive Explanation (SHAP) framework; SHAP values were used to identify relevant dosimetric thresholds and were converted to odds ratios (ORs) with confidence intervals (CIs) generated using bootstrapping to obtain quantitative measures of risk. Thresholds were validated using logistic regression.

RESULTS

The best-performing models for pulmonary, cardiac, and esophageal toxicities, outperforming logistic regression, were extreme gradient boosting (AUC, 0.739), random forest (AUC, 0.706), and naive Bayes (AUC, 0.721), respectively. For pulmonary toxicity, thresholds of a mean dose >18 Gy (OR, 2.467; 95% CI, 1.049-5.800; P = .038) and lung volume receiving ≥20 Gy (V20) > 37% (OR, 2.722; 95% CI, 1.034-7.163; P = .043) were identified. For esophageal toxicity, thresholds of a mean dose >34 Gy (OR, 4.006; 95% CI, 2.183-7.354; P < .001) and V20 > 37% (OR, 3.725; 95% CI, 1.308-10.603; P = .014) were identified. No significant thresholds were identified for cardiac toxicity.

CONCLUSIONS

In this data set, ML approaches validated known dosimetric thresholds and outperformed logistic regression at predicting toxicity. Furthermore, using explainable artificial intelligence, clinically useful dosimetric thresholds might be identified and subsequently externally validated.

Peripheral T-cell lymphoma with a T follicular-helper phenotype: A different entity? Results of the Spanish Real-T study

Nodal peripheral T-cell lymphoma (PTCL) with a T follicular helper phenotype (PTCL-TFH) is a new type of PTCL. We aimed to define its clinical characteristics and prognosis compared to PTCL not otherwise specified (PTCL-NOS) and angioimmunoblastic T-cell lymphoma (AITL). This retrospective observational study included 175 patients diagnosed with PTCL between 2008 and 2013 in 13 Spanish sites. Patient diagnosis was centrally reviewed, and patients were reclassified according to the World Health Organization (WHO) 2016 criteria: 21 patients as PTCL-NOS, 55 as AITL and 23 as PTCL-TFH. Median follow-up was 56.07 months (95% CI 38.7-73.4). Progression-free survival (PFS) and overall survival (OS) were significantly higher in patients with PTCL-TFH than in those with PTCL-NOS and AITL (PFS, 24.6 months vs. 4.6 and 7.8 months, respectively, p = 0.002; OS, 52.6 months vs. 10.0 and 19.3 months, respectively, p < 0.001). Histological diagnosis maintained an independent influence on both PFS (hazard ratio [HR] 4.1 vs. PTCL-NOS, p = 0.008; HR 2.6 vs. AITL, p = 0.047) and OS (HR 5.7 vs. PTCL-NOS, p = 0.004; HR 2.6 vs. AITL, p = 0.096), regardless of the International Prognostic Index. These results suggest that PTCL-TFH could have more favourable features and prognosis than the other PTCL subtypes, although larger series are needed to corroborate these findings.

Comparison of survival analysis approaches to modelling age at first sex among youth in Kisesa Tanzania

BACKGROUND

Many studies analyze sexual and reproductive event data using descriptive life tables. Survival analysis has better power to estimate factors associated with age at first sex (AFS), but proportional hazards models may not be right model to use. This study used accelerated failure time (AFT) models, restricted Mean Survival time model (RMST) models, with semi and non-parametric methods to assess age at first sex (AFS), factors associated with AFS, and verify underlying assumptions for each analysis.

METHODS

Self-reported sexual debut data was used from respondents 15-24 years in eight cross-sectional surveys between 1994-2016, and from adolescents' survey in an observational community study (2019-2020) in northwest Tanzania. Median AFS was estimated in each survey using non-parametric and parametric models. Cox regression, AFT parametric models (exponential, gamma, generalized gamma, Gompertz, Weibull, log-normal and log-logistic), and RMST were used to estimate and identify factors associated with AFS. The models were compared using Akaike information criterion (AIC) and Bayesian information criterion (BIC), where lower values represent a better model fit.

RESULTS

The results showed that in every survey, the Cox regression model had higher AIC and BIC compared to the other models. Overall, AFT had the best fit in every survey round. The estimated median AFS using the parametric and non-parametric methods were close. In the adolescent survey, log-logistic AFT showed that females and those attending secondary and higher education level had a longer time to first sex (Time ratio (TR) = 1.03; 95% CI: 1.01-1.06, TR = 1.05; 95% CI: 1.02-1.08, respectively) compared to males and those who reported not being in school. Cell phone ownership (TR = 0.94, 95% CI: 0.91-0.96), alcohol consumption (TR = 0.88; 95% CI: 0.84-0.93), and employed adolescents (TR = 0.95, 95% CI: 0.92-0.98) shortened time to first sex.

CONCLUSION

The AFT model is better than Cox PH model in estimating AFS among the young population.

DoSurvive: A webtool for investigating the prognostic power of a single or combined cancer biomarker

We present DoSurvive, a user-friendly survival analysis web tool and a cancer prognostic biomarker centered database. DoSurvive is the first database that allows users to perform multivariant survival analysis for cancers with customized gene/patient list. DoSurvive offers three survival analysis methods, Log rank test, Cox regression and accelerated failure time model (AFT), for users to analyze five types of quantitative features (mRNA, miRNA, lncRNA, protein and methylation of CpG islands) with four survival types, i.e. overall survival, disease-specific survival, disease-free interval, and progression-free interval, in 33 cancer types. Notably, the implemented AFT model provides an alternative method for genes/features which failed the proportional hazard assumption in Cox regression. With the unprecedented number of survival models implemented and high flexibility in analysis, DoSurvive is a unique platform for the identification of clinically relevant targets for cancer researcher and practitioners. DoSurvive is freely available at http://dosurvive.lab.nycu.edu.tw/.

Meta-epidemiological review identified variable reporting and handling of time-to-event analyses in publications of trials included in meta-analyses of systematic reviews

OBJECTIVES

Previous findings indicate limited reporting of systematic reviews with meta-analyses of time-to-event (TTE) outcomes. We assessed corresponding available information in trial publications included in such meta-analyses.

STUDY DESIGN AND SETTING

We extracted data from all randomized trials in pairwise, hazard ratio (HR)-based meta-analyses of primary outcomes and overall survival of 50 systematic reviews systematically identified from the Cochrane Database and Core Clinical Journals. Data on methods and characteristics relevant for TTE analysis of reviews, trials, and outcomes were extracted.

RESULTS

Meta-analyses included 235 trials with 315 trial analyses. Most prominently assessed was overall survival (91%). Definitions (61%), censoring reasons (41%), and follow-up specifications (56%) for trial outcomes were often missing. Available TTE data per trial were most frequently survival curves (83%), log-rank P values (76%), and HRs (72%). When trial TTE data recalculation was reported, reviews mostly specified HRs or P values (each 5%). Reviews primarily included intention-to-treat analyses (64%) and analyses not adjusted for covariates (25%). Except for missing outcome data, TTE-relevant trial characteristics, for example, informative censoring, treatment switching, and proportional hazards, were sporadically addressed in trial publications. Reporting limitations in trial publications translate to the review level.

CONCLUSION

TTE (meta)-analyses, in trial and review publications, need clear reporting standards.

Quantification of follow-up time in oncology clinical trials with a time-to-event endpoint: Asking the right questions

For the analysis of a time-to-event endpoint in a single-arm or randomized clinical trial it is generally perceived that interpretation of a given estimate of the survival function, or the comparison between two groups, hinges on some quantification of the amount of follow-up. Typically, a median of some loosely defined quantity is reported. However, whatever median is reported, is typically not answering the question(s) trialists actually have in terms of follow-up quantification. In this paper, inspired by the estimand framework, we formulate a comprehensive list of relevant scientific questions that trialists have when reporting time-to-event data. We illustrate how these questions should be answered, and that reference to an unclearly defined follow-up quantity is not needed at all. In drug development, key decisions are made based on randomized controlled trials, and we therefore also discuss relevant scientific questions not only when looking at a time-to-event endpoint in one group, but also for comparisons. We find that different thinking about some of the relevant scientific questions around follow-up is required depending on whether a proportional hazards assumption can be made or other patterns of survival functions are anticipated, for example, delayed separation, crossing survival functions, or the potential for cure. We conclude the paper with practical recommendations.

Variations in the Uptake of Active Surveillance for Prostate Cancer and Its Impact on Outcomes

Background

Regional differences in active surveillance (AS) uptake for prostate cancer (PC) illustrate an inequality in treatment strategies.

Objective

To examine the association between regional differences in AS uptake and transition to radical treatment, start of androgen deprivation therapy (ADT), watchful waiting, or death.

Design setting and participants

A Swedish population-based cohort study was conducted including men in the National Prostate Cancer Register in Sweden with low-risk or favorable intermediate-risk PC, starting AS from January 1, 2007 and continuing till December 31, 2019.

Intervention

Regional tradition of low, intermediate, or high proportions of immediate radical treatment.

Outcomes measurements and statistical analysis

Probabilities of transition from AS to radical treatment, start of ADT, watchful waiting, or death from other causes were assessed.

Results and limitations

We included 13 679 men. The median age was 66 yr, median PSA 5.1 ng/ml, and median follow-up 5.7 yr. Men from regions with a high AS uptake had a lower probability of transition to radical treatment (36%) than men from regions with a low AS uptake (40%; absolute difference 4.1%; 95% confidence interval [CI] 1.0-7.2), but not a higher probability of AS failure defined as the start of ADT (absolute difference 0.4%; 95% CI -0.7 to 1.4). There were no statistically significant differences in the probability of transition to watchful waiting or death from other causes. Limitations include uncertainty in the estimation of remaining lifetime and transition to watchful waiting.

Conclusions

A regional tradition of a high AS uptake is associated with a lower probability of transition to radical treatment, but not with AS failure. A low AS uptake suggests overtreatment.

Patient summary

There are considerable regional differences in active surveillance (AS) uptake for prostate cancer. This study compared the outcomes of AS in different regions and found no association between AS uptake and failure of AS; it suggests that a low AS uptake indicates overtreatment.

Induction therapy with hepatic arterial infusion chemotherapy enhances the efficacy of lenvatinib and pd1 inhibitors in treating hepatocellular carcinoma patients with portal vein tumor thrombosis

BACKGROUND

Hepatic arterial infusion chemotherapy (HAIC) with fluorouracil, leucovorin, and oxaliplatin (FOLFOX), lenvatinib and programmed death receptor-1 signaling inhibitors (PD1s) all alone have been proven effective in treating advanced hepatocellular carcinoma (HCC), yet the efficacy and safety of the tri-combination therapy in treating HCC patients with portal vein tumor thrombosis (PVTT) remains unknown.

METHODS

In this retrospective study, HCC patients with PVTT received either induction therapy of HAIC and lenvatinib plus PD1s in the initial period of treatment and then dual maintenance therapy of lenvatinib and PD1s (HAIC-Len-PD1) or continuous lenvatinib combined with PD1s (Len-PD1).

RESULTS

In total, 53 and 89 patients were enrolled into the Len-PD1 group and HAIC-Len-PD1 group, respectively. The median overall survival times were 13.8 months in the Len-PD1 group and 26.3 months in the HAIC-Len-PD1 group (hazard ratio (HR) = 0.43, P < 0.001). The median progression-free survival (PFS) time was significantly longer in the HAIC-Len-PD1 group than in the Len-PD1 group (11.5 months versus 5.5 months, HR = 0.43, P < 0.001). Induction therapy showed an objective response rate (ORR) 3 times higher than lenvatinib combined with PD1s therapy (61.8% versus 20.8%, P < 0.001), and exhibited inspiring intra- and extra-hepatic tumor control ability. Induction therapy led to more adverse events than lenvatinib combined with PD1s therapy, most of which were tolerable and controllable.

CONCLUSION

The induction therapy of FOLFOX-HAIC and lenvatinib plus PD1s is an effective and safe treatment for HCC patients with PVTT. The concept of induction therapy could be applied to other local-regional treatments and drugs combinations in HCC management.

Utility of gamma-glutamyl transpeptidase to lymphocyte count ratio in predicting prognosis of patients with oral cancer: A prospective cohort study in Southeastern China

BACKGROUND

To assess the prognostic role of gamma-glutamyl transpeptidase to lymphocyte count ratio (GLR) and develop a prognostic nomogram for patients with oral cancer.

METHODS

A prospective cohort (n = 1011) was conducted during July 2002 to March 2021 in Southeastern China.

RESULTS

The median follow-up time was 3.5 years. Multivariate Cox regression (OS: HR = 1.51, 95% CI: 1.04, 2.18) and Fine-Gray model (DSS: HR = 1.68, 95% CI: 1.14, 2.49) both showed that high GLR could act as an indicator of poor prognosis. A nonlinear dose-response relationship was observed between continuous GLR and the risk of all-cause mortality (p for overall = 0.028, p for nonlinear = 0.048). Compare with TNM stage, time-dependent ROC curve proved that GLR-based nomogram model performs better in predicting prognosis (the area under curve for 1-, 3-, and 5-years mortality: 0.63, 0.65, and 0.64 vs. 0.76, 0.77, and 0.78, p < 0.001).

CONCLUSION

GLR might be a useful tool in predicting prognosis for patients with oral cancer.

Survival Analysis of Patients with Colorectal Cancer Undergoing Combined Treatment: A Retrospective Cohort Study

BACKGROUND

If colorectal cancer (CRC) is diagnosed in the early stages, the patients will have higher survival rates. Although some other factors might affect the survival rate, the type of treatment available based on existing health and therapeutic facilities is extremely important as well. Accordingly, this study aimed to explore the best type of treatment for CRC patients.

STUDY DESIGN

This study employed a retrospective population-based cohort design.

METHODS

The data of 335 patients with CRC in Kurdistan province were collected through a population-based cancer registry system from March 1, 2009 to 2014. Demographic and clinical-pathologic data of the patients were gathered through their medical records, pathology reports, and reference to patients' homes. The survival rate was calculated using the Kaplan-Meier curve, log-rank test, and univariate and multivariate Cox regression. The data were analyzed using Stata 14 software.

RESULTS

In this study, the mean age±standard deviation at diagnosis was 61.7± 1.05 in men and 60.5± 1.12 in women, respectively, and 203 (60.5%) patients were males. There was less mortality rate among the patients who received both surgical and chemotherapy treatments compared to those who did not receive any treatment (Hazard ratio [HR]=0.57, 95% CI: 0.24-0.93).

CONCLUSION

When CRC patients are treated using both surgical and chemotherapy treatments, they will exhibit a higher survival rate. Therefore, it is suggested to use both treatments for CRC patients.

A risk-based postresection follow-up strategy for hepatocellular carcinoma patients

BACKGROUND

The optimal intervals for follow-up after hepatocellular carcinoma (HCC) patients undergo curative liver resection (LR) remain unclear. This study aimed to establish a risk-based post-resection follow-up strategy.

METHODS

Patients that were diagnosed with HCC and received LR from three hospitals in China were included. The risk-based strategy was established based on the random survival forest model and compared with a fixed strategy both internally and externally.

RESULTS

In total, 3447 patients from three hospitals were included. The authors' strategy showed superiority in the early detection of tumor relapse compared with fixed surveillance. Under fewer total visits, risk-based strategy achieved analogous survival time compared to the total 20 times follow-ups based on fixed strategy. Twelve total visits (five, three, one, two, and one visits in years 1-5, respectively) for American Joint Committee on Cancer/International Union Against Cancer T1a stage patients, 13 total visits (five, four, one, two, and one visits in years 1-5, respectively) for T1b stage patients, 15 total visits (eight, three, three, zero, and one visits in years 1-5, respectively) for T2 stage patients, and 15 total visits (eight, four, one, one, and one visits in years 1-5, respectively) for T3 stage patients were advocated. The detailed follow-up arrangements were available to the public through an interactive website (https://sysuccfyz.shinyapps.io/RiskBasedFollowUp/).

CONCLUSION

This risk-based surveillance strategy was demonstrated to detect relapse earlier and reduce the total number of follow-ups without compromising on survival. Based on the strategy and methodology of the authors, surgeons or patients could choose more intensive or flexible schedules depending on the requirements and economic conditions.

PLAIN LANGUAGE SUMMARY

A risk-based post-resection follow-up strategy was established by random survival forest model using a larger hepatocellular carcinoma population The strategy was demonstrated to detect tumor relapse earlier and reduce the total number of follow-ups without compromising on survival Our strategy and methodology could be widely applied by other surgeons and patients.

Time without PSA recurrence after radical prostatectomy as a predictor of future biochemical recurrence, metastatic disease and prostate cancer death: a prospective Scandinavian cohort study

OBJECTIVE

Although surveillance after radical prostatectomy routinely includes repeated prostate specific antigen (PSA)-testing for many years, biochemical recurrence often occurs without further clinical progression. We therefore hypothesised that follow-up can be shortened for many patients without increasing the risk of prostate cancer death. We investigated the long-term probabilities of PSA recurrence, metastases and prostate cancer death in patients without biochemical recurrence five and 10 years after radical prostatectomy.

DESIGN

Prospective cohort study. Stratification by Gleason score (≤3+4=7 or ≥4+3=7), pathological tumour stage (pT2 or ≥pT3) and negative or positive surgical margins.

SETTING

Between 1989 and 1998, 14 urological centres in Scandinavia randomised patients to the Scandinavian Prostate Cancer Group study number 4 (SPCG-4) trial.

PARTICIPATION

All 306 patients from the SPCG-4 trial who underwent radical prostatectomy within 1 year from inclusion were eligible. Four patients were excluded due to surgery-related death (n=1) or salvage radiotherapy or hormonal treatment within 6 weeks from surgery (n=3).

PRIMARY OUTCOME MEASURES

Cumulative incidences and absolute differences in metastatic disease and prostate cancer death.

RESULTS

We analysed 302 patients with complete follow-up during a median of 24 years. Median preoperative PSA was 9.8 ng/mL and median age was 65 years. For patients without biochemical recurrence 5 years after radical prostatectomy the 20-year probability of biochemical recurrence was 25% among men with Gleason score ≤3+4=7 and 57% among men with Gleason score ≥4+3=7; the probabilities for metastases were 0.8% and 17%; and for prostate cancer death 0.8% and 12%, respectively. The long-term probabilities were higher for pT ≥3 versus pT2 and for positive versus negative surgical margins. Limitations include small size of the cohort.

CONCLUSION

Many patients with favourable histopathology without biochemical recurrence 5 years after radical prostatectomy could stop follow-up earlier than 10 years after radical prostatectomy.

Machine learning models for identifying predictors of clinical outcomes with first-line immune checkpoint inhibitor therapy in advanced non-small cell lung cancer

Immune checkpoint inhibitors (ICIs) are standard-of-care as first-line (1L) therapy for advanced non-small cell lung cancer (aNSCLC) without actionable oncogenic driver mutations. While clinical trials demonstrated benefits of ICIs over chemotherapy, variation in outcomes across patients has been observed and trial populations may not be representative of clinical practice. Predictive models can help understand heterogeneity of treatment effects, identify predictors of meaningful clinical outcomes, and may inform treatment decisions. We applied machine learning (ML)-based survival models to a real-world cohort of patients with aNSCLC who received 1L ICI therapy extracted from a US-based electronic health record database. Model performance was evaluated using metrics including concordance index (c-index), and we used explainability techniques to identify significant predictors of overall survival (OS) and progression-free survival (PFS). The ML model achieved c-indices of 0.672 and 0.612 for OS and PFS, respectively, and Kaplan-Meier survival curves showed significant differences between low- and high-risk groups for OS and PFS (both log-rank test p < 0.0001). Identified predictors were mostly consistent with the published literature and/or clinical expectations and largely overlapped for OS and PFS; Eastern Cooperative Oncology Group performance status, programmed cell death-ligand 1 expression levels, and serum albumin were among the top 5 predictors for both outcomes. Prospective and independent data set evaluation is required to confirm these results.

Examining evidence of time-dependent treatment effects: an illustration using regression methods

BACKGROUND

For the design and analysis of clinical trials with time-to-event outcomes, the Cox proportional hazards model and the logrank test have been the cornerstone methods for many decades. Increasingly, the key assumption of proportionality-or time-fixed effects-that underpins these methods has been called into question. The availability of novel therapies with new mechanisms of action and clinical trials of longer duration mean that non-proportional hazards are now more frequently encountered.

METHODS

We compared several regression-based methods to model time-dependent treatment effects. For illustration purposes, we used selected endpoints from a large, community-based clinical trial of low dose daily aspirin in older persons. Relative and absolute estimands were defined, and analyses were conducted in all participants. Additional exploratory analyses were undertaken by selected subgroups of interest using interaction terms in the regression models.

DISCUSSION

In the trial with median 4.7 years follow-up, we found evidence for non-proportionality and a time-dependent treatment effect of aspirin on cancer mortality not previously reported in trial findings. We also found some evidence of time-dependence to an aspirin by age interaction for major adverse cardiovascular events. For other endpoints, time-fixed treatment effect estimates were confirmed as appropriate.

CONCLUSIONS

The consideration of treatment effects using both absolute and relative estimands enhanced clinical insights into potential dynamic treatment effects. We recommend these analytical approaches as an adjunct to primary analyses to fully explore findings from clinical trials.

Efficient and accurate frailty model approach for genome-wide survival association analysis in large-scale biobanks

With decades of electronic health records linked to genetic data, large biobanks provide unprecedented opportunities for systematically understanding the genetics of the natural history of complex diseases. Genome-wide survival association analysis can identify genetic variants associated with ages of onset, disease progression and lifespan. We propose an efficient and accurate frailty model approach for genome-wide survival association analysis of censored time-to-event (TTE) phenotypes by accounting for both population structure and relatedness. Our method utilizes state-of-the-art optimization strategies to reduce the computational cost. The saddlepoint approximation is used to allow for analysis of heavily censored phenotypes (>90%) and low frequency variants (down to minor allele count 20). We demonstrate the performance of our method through extensive simulation studies and analysis of five TTE phenotypes, including lifespan, with heavy censoring rates (90.9% to 99.8%) on ~400,000 UK Biobank participants with white British ancestry and ~180,000 individuals in FinnGen. We further analyzed 871 TTE phenotypes in the UK Biobank and presented the genome-wide scale phenome-wide association results with the PheWeb browser.

Ultra-radical (extensive) surgery versus standard surgery for the primary cytoreduction of advanced epithelial ovarian cancer

BACKGROUND

Ovarian cancer is the seventh most common cancer among women and the leading cause of death in women with gynaecological malignancies. Opinions differ regarding the role of ultra-radical (extensive) cytoreductive surgery in ovarian cancer treatment.

OBJECTIVES

To evaluate the effectiveness and morbidity associated with ultra-radical/extensive surgery in the management of advanced-stage epithelial ovarian cancer.

SEARCH METHODS

We searched CENTRAL (2021, Issue 11), MEDLINE Ovid and Embase Ovid up to November 2021. We also searched registers of clinical trials, abstracts of scientific meetings, reference lists of included studies and contacted experts in the field.

SELECTION CRITERIA

Randomised controlled trials (RCTs) or non-randomised studies (NRS), analysed using multivariate methods, that compared ultra-radical/extensive and standard surgery in women with advanced primary epithelial ovarian cancer.

DATA COLLECTION AND ANALYSIS

Two review authors independently assessed whether potentially relevant studies met the inclusion criteria, abstracted data and assessed the risk of bias. We identified three NRS and conducted meta-analyses where possible.

MAIN RESULTS

We identified three retrospective observational studies for inclusion in the review. Two studies included women exclusively undergoing upfront primary debulking surgery (PDS) and the other study including both PDS and interval debulking surgical (IDS) procedures. All studies were at critical risk of bias due to retrospective and non-randomised study designs. Meta-analysis of two studies, assessing 397 participants, found that women who underwent radical procedures, as part of PDS, may have a lower risk of mortality compared to women who underwent standard surgery (adjusted HR 0.60, 95% CI 0.43 to 0.82; I2 = 0%; very low-certainty evidence), but the evidence is very uncertain. The results were robust to a sensitivity analysis including women with more-extensive disease (carcinomatosis) (adjusted HR 0.61, 95% CI 0.44 to 0.85; I2 = 0%; n = 283, very low-certainty evidence), but the evidence is very uncertain. One study reported a comparison of radical versus standard surgical procedures associated with both PDS and IDS procedures, but a multivariate analysis was only undertaken for disease-free survival (DFS) and therefore the certainty of the evidence was not assessable for overall survival (OS) and remains very low. The lack of reporting of OS meant the study was at high risk of bias for selective reporting of outcomes. One study, 203 participants, found that women who underwent radical procedures as part of PDS may have a lower risk of disease progression or death compared to women who underwent standard surgery (adjusted HR 0.62, 95% CI 0.42 to 0.92; very low-certainty evidence), but the evidence is very uncertain. The results were robust to a sensitivity analysis in one study including women with carcinomatosis (adjusted HR 0.52, 95% CI 0.33 to 0.82; n = 139; very low-certainty evidence), but the evidence is very uncertain. A combined analysis in one study found that women who underwent radical procedures (using both PDS and IDS) may have an increased chance of disease progression or death than those who received standard surgery (adjusted HR 1.60, 95% CI 1.11 to 2.31; I2 = 0%; n = 527; very low-certainty evidence), but the evidence is very uncertain. In absolute and unadjusted terms, the DFS was 19.3 months in the standard surgery group, 15.8 in the PDS group and 15.9 months in the IDS group. All studies were at critical risk of bias and we only identified very low-certainty evidence for all outcomes reported in the review. Perioperative mortality, adverse events and quality of life (QoL) outcomes were either not reported or inadequately reported in the included studies. Two studies reported perioperative mortality (death within 30 days of surgery), but they did not use any statistical adjustment. In total, there were only four deaths within 30 days of surgery in both studies. All were observed in the standard surgery group, but we did not report a risk ratio (RR) to avoid potentially misleading results with so few deaths and very low-certainty evidence. Similarly, one study reported postoperative morbidity, but the authors did not use any statistical adjustment. Postoperative morbidity occurred more commonly in women who received ultra-radical surgery compared to standard surgery, but the certainty of the evidence was very low.

AUTHORS' CONCLUSIONS

We found only very low-certainty evidence comparing ultra-radical surgery and standard surgery in women with advanced ovarian cancer. The evidence was limited to retrospective, NRSs and so is at critical risk of bias. The results may suggest that ultra-radical surgery could result in improved OS, but results are based on very few women who were chosen to undergo each intervention, rather than a randomised study and intention-to-treat analysis, and so the evidence is very uncertain. Results for progression/DFS were inconsistent and evidence was sparse. QoL and morbidity was incompletely or not reported in the three included studies. A separate prognostic review assessing residual disease as a prognostic factor in this area has been addressed elsewhere, which demonstrates the prognostic effect of macroscopic debulking to no macroscopic residual disease. In order to aid existing guidelines, the role of ultra-radical surgery in the management of advanced-stage ovarian cancer could be addressed through the conduct of a sufficiently powered, RCT comparing ultra-radical and standard surgery, or well-designed NRSs, if this is not possible.

dsSurvival: Privacy preserving survival models for federated individual patient meta-analysis in DataSHIELD

Objective

Achieving sufficient statistical power in a survival analysis usually requires large amounts of data from different sites. Sensitivity of individual-level data, ethical and practical considerations regarding data sharing across institutions could be a potential challenge for achieving this added power. Hence we implemented a federated meta-analysis approach of survival models in DataSHIELD, where only anonymous aggregated data are shared across institutions, while simultaneously allowing for exploratory, interactive modelling. In this case, meta-analysis techniques to combine analysis results from each site are a solution, but an analytic workflow involving local analysis undertaken at individual studies hinders exploration. Thus, the aim is to provide a framework for performing meta-analysis of Cox regression models across institutions without manual analysis steps for the data providers.

Results

We introduce a package (dsSurvival) which allows privacy preserving meta-analysis of survival models, including the calculation of hazard ratios. Our tool can be of great use in biomedical research where there is a need for building survival models and there are privacy concerns about sharing data.

Kidney Graft Failure and Patient Survival Modelling Based on Competing Risks Under Nonproportional Hazards

We analyze data on Silesian patients after kidney transplantation under competing events scenarios where time to death and time to graft failure are considered as absorbing competing events. Our objectives are to use model diagnostics in identifying violations of proportionality assumption under the framework of subdistribution and cause-specific hazards. We use the Fine-Gray proportional hazards model for the subdistribution. Under the cause-specific hazards (CSH) scenario we use the Cox proportional hazards model and Gray's time-varying coefficients model and available model diagnostics. We show that violation of proportional subdistribution hazards assumption may be conveniently identified using residual diagnostics and properly accounted for by involving time interactions with appropriate model predictors. We also show that although the nonproportional effects on cumulative incidence do not necessarily translate in those on cause-specific hazards, they often take place simultaneously, and a violation of the proportionality assumption needs to be checked rigorously. Time-varying effects have a profound impact on clinical inference under competing risks. They do not translate directly between the frameworks of subdistribution and cause-specific hazards because the cumulative incidence is obtained via integrating the cause-specific hazard weighted by the overall survival function. Also, a different definition of the risk set is in place under the cumulative incidence and CSH framework, respectively. However, a simultaneous violation of the proportionality assumption under both frameworks is still possible. Clinical inference may change considerably when such a violation occurs. Nonproportional effects may be properly identified under each framework using available model diagnostics.

Violations of proportional hazard assumption in Cox regression model of transcriptomic data in TCGA pan-cancer cohorts

Background

Cox proportional hazard regression (CPH) model relies on the proportional hazard (PH) assumption: the hazard of variables is independent of time. CPH has been widely used to identify prognostic markers of the transcriptome. However, the comprehensive investigation on PH assumption in transcriptomic data has lacked.

Results

The whole transcriptomic data of the 9,056 patients from 32 cohorts of The Cancer Genome Atlas and the 3 lung cancer cohorts from Gene Expression Omnibus were collected to construct CPH model for each gene separately for fitting the overall survival. An average of 8.5% gene CPH models violated the PH assumption in TCGA pan-cancer cohorts. In the gene interaction networks, both hub and non-hub genes in CPH models were likely to have non-proportional hazards. Violations of PH assumption for the same gene models were not consistent in 5 non-small cell lung cancer datasets (all kappa coefficients < 0.2), indicating that the non-proportionality of gene CPH models depended on the datasets. Furthermore, the introduction of log(t) or sqrt(t) time-functions into CPH improved the performance of gene models on overall survival fitting in most tumors. The time-dependent CPH changed the significance of log hazard ratio of the 31.9% gene variables.

Conclusions

Our analysis resulted that non-proportional hazards should not be ignored in transcriptomic data. Introducing time interaction term ameliorated performance and interpretability of non-proportional hazards of transcriptome data in CPH.

Metastatic uveal melanoma: The final frontier

Treatment of primary intraocular uveal melanoma has developed considerably, its driver genes are largely unraveled, and the ways to assess its risk for metastases are very precise, being based on an international staging system and genetic data. Unfortunately, the risk of distant metastases, which emerge in approximately one half of all patients, is unaltered. Metastases are the leading single cause of death after uveal melanoma is diagnosed, yet no consensus exists regarding surveillance, staging, and treatment of disseminated disease, and survival has not improved until recently. The final frontier in conquering uveal melanoma lies in solving these issues to cure metastatic disease. Most studies on metastatic uveal melanoma are small, uncontrolled, retrospective, and do not report staging. Meta-analyses confirm a median overall survival of 10-13 months, and a cure rate that approaches nil, although survival exceeding 5 years is possible, estimated 2% either with first-line treatment or with best supportive care. Hepatic ultrasonography and magnetic resonance imaging as surveillance methods have a sensitivity of 95-100% and 83-100%, respectively, to detect metastases without radiation hazard according to prevailing evidence, but computed tomography is necessary for staging. No blood-based tests additional to liver function tests are generally accepted. Three validated staging systems predict, each in defined situations, overall survival after metastasis. Their essential components include measures of tumor burden, liver function, and performance status or metastasis free interval. Age and gender may additionally influence survival. Exceptional mutational events in metastases may make them susceptible to checkpoint inhibitors. In a large meta-analysis, surgical treatment was associated with 6 months longer median overall survival as compared to conventional chemotherapy and, recently, tebentafusp as first-line treatment at the first interim analysis of a randomized phase III trial likewise provided a 6 months longer median overall survival compared to investigator's choice, mostly pembrolizumab; these treatments currently apply to selected patients. Promoting dormancy of micrometastases, harmonizing surveillance protocols, promoting staging, identifying predictive factors, initiating controlled clinical trials, and standardizing reporting will be critical steppingstones in reaching the final frontier of curing metastatic uveal melanoma.

Comprehensive Statistical Exploration of Prognostic (Bio-)Markers for Responses to Immune Checkpoint Inhibitor in Patients with Non-Small Cell Lung Cancer

Simple Summary

Metastatic non-small cell lung cancer (NSCLC) patients treated with immune checkpoint inhibitors (ICIs) may suffer from heavy side effects, and not all patients benefit from the treatment. Therefore, it is crucial to gain knowledge about possible (bio-)markers for response to ICIs. We used retrospective data acquired from NSCLC patients treated with ICIs in first- or further-line therapy settings, including 16 possible markers. We conducted a comprehensive statistical analysis study to find markers for response to treatment, assessed the robustness of our results, and discussed often encountered statistical pitfalls. Our study yielded hypotheses for various predictive and prognostic (bio-)markers for response to ICIs in NSCLC patients. In particular, we found that high basophil counts may be predictive for treatment response in patients in further-line therapy settings.

Abstract

Metastatic non-small cell lung cancer (NSCLC) patients treated with immune checkpoint inhibitors (ICIs) may suffer from heavy side effects and not all patients benefit from the treatment. We conducted a comprehensive statistical analysis to identify promising (bio-)markers for treatment response. We analyzed retrospective data from NSCLC patients treated with ICIs in first- or further-line therapy settings at the University Hospital Zurich. We investigated 16 possible prognostic markers with respect to overall survival, tumor size reduction, and the development of an immune-related adverse event (irAE) and assessed the robustness of our results. For the further-line patient group, the most significant result was that increased basophil counts were associated with increased odds of tumor size reduction within three months and with the development of an irAE. For the first-line patient group, the most significant results were that increased lymphocyte counts, the histology of adenocarcinoma, and the intake of non-steroidal anti-rheumatic drugs (NSAR) were associated with decreased hazards of dying. Our study yielded new hypotheses for predictive (bio-)markers for response to ICIs in NSCLC patients. The possibly beneficial role of high basophil counts is a particularly interesting finding. Our results should be tested on independent data in a prospective fashion.

Prognostic value of baseline metabolic tumour volume in advanced-stage Hodgkin's lymphoma

Our aim was to evaluate the prognostic value of initial total metabolic tumour volume (TMTV) in a population of patients with advanced-stage Hodgkin's lymphoma (HL). We retrospectively included 179 patients with stage IIb-III-IV Hodgkin's disease who received BEACOPP or ABVD as the first-line treatment. The initial TMTV was determined using a semi-automatic method for each patient. We analysed its prognostic value in terms of 5-year progression-free survival (PFS), overall survival, and positron emission tomography (PET) response after two courses of chemotherapy. Considering all the treatments and using a threshold of 217 cm³, TMTV was predictive of 5-year PFS and PET response after two courses of chemotherapy. In multivariable analysis involving TMTV, IPI score, and the first treatment received, TMTV remained a baseline prognostic factor for 5-year PFS. In the subgroup of patients treated with BEACOPP with a threshold of 331 cm³, TMTV was predictive of PET response, but not 5-year PFS (p = 0.087). The combined analysis of TMTV and PET response enabled the individualisation of a subgroup of patients (low TMTV and complete response on PET) with a very low risk of recurrence. Baseline TMTV appears to be a useful independent prognostic factor for predicting relapse in advanced-stage HL in ABVD subgroup, with a tendency of survival curves separation in BEACOPP subgroup.

Using fractional polynomials and restricted cubic splines to model non-proportional hazards or time-varying covariate effects in the Cox regression model

The Cox proportional hazards model is used extensively in clinical and epidemiological research. A key assumption of this model is that of proportional hazards. A variable satisfies the proportional hazards assumption if the effect of that variable on the hazard function is constant over time. When the proportional hazards assumption is violated for a given variable, a common approach is to modify the model so that the regression coefficient associated with the given variable is assumed to be a linear function of time (or of log‐time), rather than being constant or fixed. However, this is an unnecessarily restrictive assumption. We describe two different methods to allow a regression coefficient, and thus the hazard ratio, in a Cox model to vary as a flexible function of time. These methods use either fractional polynomials or restricted cubic splines to model the log‐hazard ratio as a function of time. We illustrate the utility of these methods using data on 12 705 patients who presented to a hospital emergency department with a primary diagnosis of heart failure. We used a Cox model to assess the association between elevated cardiac troponin at presentation and the hazard of death after adjustment for an extensive set of covariates. SAS code for implementing the restricted cubic spline approach is provided, while an existing Stata function allows for the use of fractional polynomials.

Using median survival in meta-analysis of experimental time-to-event data

BACKGROUND

Time-to-event data is frequently reported in both clinical and preclinical research spheres. Systematic review and meta-analysis is a tool that can help to identify pitfalls in preclinical research conduct and reporting that can help to improve translational efficacy. However, pooling of studies using hazard ratios (HRs) is cumbersome especially in preclinical meta-analyses including large numbers of small studies. Median survival is a much simpler metric although because of some limitations, which may not apply to preclinical data, it is generally not used in survival meta-analysis. We aimed to appraise its performance when compared with hazard ratio-based meta-analysis when pooling large numbers of small, imprecise studies.

METHODS

We simulated a survival dataset with features representative of a typical preclinical survival meta-analysis, including with influence of a treatment and a number of covariates. We calculated individual patient data-based hazard ratios and median survival ratios (MSRs), comparing the summary statistics directly and their performance at random-effects meta-analysis. Finally, we compared their sensitivity to detect associations between treatment and influential covariates at meta-regression.

RESULTS

There was an imperfect correlation between MSR and HR, although the opposing direction of treatment effects between summary statistics appeared not to be a major issue. Precision was more conservative for HR than MSR, meaning that estimates of heterogeneity were lower. There was a slight sensitivity advantage for MSR at meta-analysis and meta-regression, although power was low in all circumstances.

CONCLUSIONS

We believe we have validated MSR as a summary statistic for use in a meta-analysis of small, imprecise experimental survival studies-helping to increase confidence and efficiency in future reviews in this area. While assessment of study precision and therefore weighting is less reliable, MSR appears to perform favourably during meta-analysis. Sensitivity of meta-regression was low for this set of parameters, so pooling of treatments to increase sample size may be required to ensure confidence in preclinical survival meta-regressions.

Avelumab in newly diagnosed glioblastoma

Background

Glioblastoma (GBM) is known to use both local and systemic immunosuppressive strategies. One such strategy is the expression of the immune checkpoint protein programmed cell death ligand-1 (PD-L1) by both tumor cells and tumor-associated immune cells. Recent phase III trials using IgG4 antibodies targeting PD-1, the ligand for PD-L1, failed to show any benefit. Avelumab is an IgG1 monoclonal antibody targeting PD-L1. In contrast to the previously tested immune checkpoint inhibitors, it can directly bind tumor cells and immune cells expressing PD-L1 and can induce antibody-dependent cellular cytotoxicity.

Methods

We conducted a single center, open label, phase II study where avelumab 10 mg/kg IV Q2W was added concurrently to the first monthly temozolomide cycle in patients with newly diagnosed GBM. Immunohistochemical analyses were performed on surgery samples. The primary objective was safety. Secondary objectives were efficacy outcomes according to the immunotherapy Response Assessment in Neuro Oncology criteria, progression free survival (PFS), and overall survival (OS). Exploratory objectives aimed at determining prognostic biomarkers.

Results

Thirty patients were started on therapy and two were lost to follow-up. Median follow-up time (reverse Kaplan-Meier) was 41.7 months (IQR: 28.3–43.4). Three (10.0%) patients had a related or possibly related treatment emergent adverse event that lead to transient or permanent discontinuation of avelumab. Eight (26.7%) patients had one or more immune-related adverse events, and 8 (26.7%) patients had an infusion-related reaction. The overall response rate was 23.3%, median PFS was 9.7 months, and the median OS was 15.3 months. No pretreatment biomarkers showed any predictive value.

Conclusions

The addition of avelumab to standard therapy in patients with GBM was not associated with any new safety signal. There was no apparent improvement in OS.

Trial Registration

NCT03047473 Registered February 9, 2017.

Practical Considerations and Challenges When Conducting an Individual Participant Data (IPD) Meta-Analysis

This chapter provides a broad overview of the use of individual participant (sometimes referred to as patient) data (IPD ) within meta-analyses, the associated advantages of using IPD in meta-analysis compared to aggregate data, and when IPD should be used in meta-analysis.This chapter also outlines the steps of conducting an IPD meta-analysis, with practical guidance relating to requesting and obtaining IPD for meta-analysis. Challenges that can be associated with conducting an IPD meta-analysis are also discussed, including consideration of availability bias, when a subset of the relevant IPD is not available for meta-analysis.

Predictive and prognostic impact of ferroptosis-related genes ACSL4 and GPX4 on breast cancer treated with neoadjuvant chemotherapy

BACKGROUND

Recent evidence shows that inducing ferroptosis may improve efficacy of tumor therapy. However, ferroptosis-related genes have been little studied in patients with breast cancer especially in the neoadjuvant setting. ACSL4 and GPX4 have been well established as the positive and negative regulator of ferroptosis, respectively. This study aimed to explore the predictive value of ACSL4 and GPX4 for patients with breast cancer administered neoadjuvant chemotherapy.

METHODS

This study included patients treated with paclitaxel-cisplatin-based neoadjuvant chemotherapy. Immunohistochemistry staining of ACSL4 and GPX4 was carried out on the core needle biopsy specimens. Logistic regression was performed to explore the predictive biomarkers of pathological complete response (pCR). Survival analyses were examined by log-rank test and Cox proportional hazard regression.

FINDINGS

A total of 199 patients were included for the analyses. Both ACSL4 expression and ACSL4/GPX4 combination status could serve as independent predictive factors for pCR. The interaction for pCR was observed between ACSL4 and clinical tumor stage. Besides, ACSL4 expression, GPX4 expression, and their combination status were independent prognostic factors for disease-free survival. Analyses of the Kaplan-Meier Plotter database suggested that higher ACSL4 expression is related to better overall survival, and higher GPX4 expression is related to better distant metastasis-free survival. Pathway analyses revealed that ACSL4 and GPX4 might function in crucial pathways including apoptosis, autophagy, cell adhesion, lipid metabolism, etc. INTERPRETATION: This study revealed the critical value of ACSL4 and GPX4 serving as novel predictive and prognostic biomarkers for patients with breast cancer receiving neoadjuvant chemotherapy. It might be a novel strategy to induce ferroptosis to promote chemosensitivity. Future studies are required to elucidate the potential mechanisms.

FUNDING

This work was supported by Shanghai Natural Science Foundation [grant number 19ZR1431100], Clinical Research Plan of Shanghai Hospital Development Center [grant numbers SHDC2020CR3003A, 16CR3065B, and 12016231], Shanghai "Rising Stars of Medical Talent" Youth Development Program for Youth Medical Talents - Specialist Program [grant number 2018-15], Shanghai "Rising Stars of Medical Talent" Youth Development Program for Outstanding Youth Medical Talents [grant number 2018-16], Shanghai Collaborative Innovation Center for Translational Medicine [grant number TM201908], Multidisciplinary Cross Research Foundation of Shanghai Jiao Tong University [grant numbers YG2017QN49, ZH2018QNA42, and YG2019QNA28], Nurturing Fund of Renji Hospital [grant numbers PYMDT-002, PY2018-IIC-01, PY2018-III-15, and PYIII20-09], Science and Technology Commission of Shanghai Municipality [grant numbers 20DZ2201600 and 15JC1402700], and Shanghai Municipal Key Clinical Specialty.

Multicenter Privacy-Preserving Cox Analysis Based on Homomorphic Encryption

The Cox proportional hazards model is one of the most widely used methods for analyzing survival data. Data from multiple data providers are required to improve the generalizability and confidence of the results of Cox analysis; however, such data sharing may result in leakage of sensitive information, leading to financial fraud, social discrimination or unauthorized data abuse. Some privacy-preserving Cox regression protocols have been proposed in past years, but they lack either security or functionality. In this paper, we propose a privacy-preserving Cox regression protocol for multiple data providers and researchers. The proposed protocol allows researchers to train models on horizontally or vertically partitioned datasets while providing privacy protection for both the sensitive data and the trained models. Our protocol utilizes threshold homomorphic encryption to guarantee security. Experimental results demonstrate that with the proposed protocol, Cox regression model training over 9 variables in a dataset of 113,035 samples takes approximately 44 min, and the trained model is almost the same as that obtained with the original nonsecure Cox regression protocol; therefore, our protocol is a potential candidate for practical real-world applications in multicenter medical research.

Machine Learning-Based Interpretation and Visualization of Nonlinear Interactions in Prostate Cancer Survival

PURPOSE

Shapley additive explanation (SHAP) values represent a unified approach to interpreting predictions made by complex machine learning (ML) models, with superior consistency and accuracy compared with prior methods. We describe a novel application of SHAP values to the prediction of mortality risk in prostate cancer.

METHODS

Patients with nonmetastatic, node-negative prostate cancer, diagnosed between 2004 and 2015, were identified using the National Cancer Database. Model features were specified a priori: age, prostate-specific antigen (PSA), Gleason score, percent positive cores (PPC), comorbidity score, and clinical T stage. We trained a gradient-boosted tree model and applied SHAP values to model predictions. Open-source libraries in Python 3.7 were used for all analyses.

RESULTS

We identified 372,808 patients meeting the inclusion criteria. When analyzing the interaction between PSA and Gleason score, we demonstrated consistency with the literature using the example of low-PSA, high-Gleason prostate cancer, recently identified as a unique entity with a poor prognosis. When analyzing the PPC-Gleason score interaction, we identified a novel finding of stronger interaction effects in patients with Gleason ≥ 8 disease compared with Gleason 6-7 disease, particularly with PPC ≥ 50%. Subsequent confirmatory linear analyses supported this finding: 5-year overall survival in Gleason ≥ 8 patients was 87.7% with PPC < 50% versus 77.2% with PPC ≥ 50% (P < .001), compared with 89.1% versus 86.0% in Gleason 7 patients (P < .001), with a significant interaction term between PPC ≥ 50% and Gleason ≥ 8 (P < .001).

CONCLUSION

We describe a novel application of SHAP values for modeling and visualizing nonlinear interaction effects in prostate cancer. This ML-based approach is a promising technique with the potential to meaningfully improve risk stratification and staging systems.

Impact of a non-constant baseline hazard on detection of time-dependent treatment effects: a simulation study

Background

Non-proportional hazards are common with time-to-event data but the majority of randomised clinical trials (RCTs) are designed and analysed using approaches which assume the treatment effect follows proportional hazards (PH). Recent advances in oncology treatments have identified two forms of non-PH of particular importance - a time lag until treatment becomes effective, and an early effect of treatment that ceases after a period of time. In sample size calculations for treatment effects on time-to-event outcomes where information is based on the number of events rather than the number of participants, there is crucial importance in correct specification of the baseline hazard rate amongst other considerations. Under PH, the shape of the baseline hazard has no effect on the resultant power and magnitude of treatment effects using standard analytical approaches. However, in a non-PH context the appropriateness of analytical approaches can depend on the shape of the underlying hazard.

Methods

A simulation study was undertaken to assess the impact of clinically plausible non-constant baseline hazard rates on the power, magnitude and coverage of commonly utilized regression-based measures of treatment effect and tests of survival curve difference for these two forms of non-PH used in RCTs with time-to-event outcomes.

Results

In the presence of even mild departures from PH, the power, average treatment effect size and coverage were adversely affected. Depending on the nature of the non-proportionality, non-constant event rates could further exacerbate or somewhat ameliorate the losses in power, treatment effect magnitude and coverage observed. No single summary measure of treatment effect was able to adequately describe the full extent of a potentially time-limited treatment benefit whilst maintaining power at nominal levels.

Conclusions

Our results show the increased importance of considering plausible potentially non-constant event rates when non-proportionality of treatment effects could be anticipated. In planning clinical trials with the potential for non-PH, even modest departures from an assumed constant baseline hazard could appreciably impact the power to detect treatment effects depending on the nature of the non-PH. Comprehensive analysis plans may be required to accommodate the description of time-dependent treatment effects.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01372-0.

Prediction Model of Lymph Node Metastasis Risk in Elderly Patients with Early Gastric Cancer before Endoscopic Resection: A Retrospective Analysis Based on International Multicenter Data

Background: Patients with early gastric cancer (EGC) must suffer reoperation if diagnosed with a high possibility of lymph node (LN) metastasis. The purpose of the current study was to develop and validate a model to predict the risk of LN metastasis in elderly patients before endoscopic resection.

Methods: A total of 1911 EGC patients who had undergone radical surgery were selected and assigned randomly (2:1) to either the training cohort or the validation cohort. A nomogram was established based on the univariate and multivariate logistic regression models using the training cohort. Cox proportional hazards regression models were applied to identify the prognostic factors in univariate and multivariable analyses.

Results: Three variables—tumor size, grade, and T stage—were derived from the multivariate analyses in the training cohort and incorporated into the nomogram. The AUC of the nomogram was 0.732 in the training cohort and 0.706 in the validation cohort. There were significant differences in survival among patients with different degrees of LN metastasis risk (training cohort: five-year disease-specific survival (DSS): low risk 88.1% vs. moderate risk 80.0% vs. high risk 72.9%, P < 0.001; validation cohort five-year DSS: low risk 89.0% vs. moderate risk 84.3% vs. high risk 72.2%, P < 0.001). The LN metastasis risk assessed from the model was also an independent prognostic factor.

Conclusion: We established a nomogram that accurately predicts LN metastasis risk for elderly patients with EGC before endoscopic resection to avoid further injury from reoperation.

Immune checkpoint inhibitors at any treatment line in advanced NSCLC: Real-world overall survival in a large Italian cohort

OBJECTIVES

To estimate the average treatment effect of immune checkpoint inhibitors in any line of treatment in a 2016-2018 population-based cohort of patients with advanced non-small-cell lung cancer (NSCLC).

MATERIALS AND METHODS

The cohort, and information on the tumor, were derived from the cancer registry of the Agency for Health Protection of Milan, Italy. Inclusion criteria were adult age, microscopically confirmed NSCLC, stage IIIB or IV at diagnosis, and having received at least one line of treatment. Treatment with all licensed anti PD-1/PD-L1 inhibitors was derived from inpatients and outpatients' pharmaceutical databases of the ATS and vital status at 31 December 2019 from the health registry office of the Lombardy region. We investigated, with a causal approach, the relationship between survival and anti PD-1/PD-L1 treatment at any line constructing a directed acyclic graph and fitting a Marginal Structural Cox Model (MSCM).

RESULTS

Of 1673 subjects, 324 received anti PD-1/PD-L1 at any treatment line. Overall, one-year survival was 61.1% (95 %CI, 55.6-66.2%) in the group treated with anti PD-1/PD-L1 at any line and 31.1% (95 %CI, 28.6-33.5%) among not treated. One-year hazard ratio (HR) of death for not treated vs. treated was 2.15 (95 %CI, 1.91-2.41), decreasing to 1.23 (95 %CI, 1.03-1.46) at two years and reaching one in the third year.

CONCLUSION

In un unselected population-based cohort with advanced lung cancer, treatment with anti PD-1/PD-L1 at any line lowered the hazard of death up to two-years from date of diagnosis, confirming the efficacy of immunotherapy outside clinical trials.

Informative censoring of surrogate end-point data in phase 3 oncology trials

BACKGROUND

Kaplan-Meier (K-M) analysis, the cornerstone of cancer clinical trial interpretation, assumes that censored patients are no more or less likely to experience an event than those followed. We sought to investigate the patterns of censoring in surrogate end-points of oncology randomised controlled trials (RCTs) and examine the relationship between censoring in practice-changing treatments that failed to demonstrate survival gain.

METHODS

In this cross-sectional study of phase III RCTs published in the New England Journal of Medicine, Lancet, and JAMA, between 2010 and 2020, K-M curves of surrogate end-points with statistical significance were extracted. The reverse K-M method (i.e., events and censoring are flipped) was used to examine differential censoring using the analogous reverse hazard ratio and restricted mean survival time. Sensitivity analysis was performed by partially restoring the balance in censoring between study arms.

RESULTS

Of the 73 eligible studies with significant surrogates, 33 (45%) reported significant overall survival benefit (concordant trials), and 40 (55%) did not (discordant trials). The proportion of studies with significant differential censoring in surrogates was 43% (17/40) and 51% (17/33) in discordant and concordant trials, respectively. Trials with a significant censoring imbalance in the experimental arm occurred only in discordant trials (15% vs 0%, odds ratio [OR] = 12.62, P = 0.033), compared to excessive censoring in the control arm which occurred more in concordant trials (28% vs 52%; OR = 0.36, P = 0.036). Although censoring imbalance occurred in both groups, after sensitivity analysis, 50% of the discordant trials lost their statistical significance, compared to 15% of concordant trials (OR = 5.6, P = 0.0018).

CONCLUSION

Censoring imbalance between study arms of RCTs suggests a potential systemic bias and raises uncertainty regarding the validity of the results. Informative censoring may explain the inconsistency between therapies that seem to improve disease outcomes without concomitant survival benefit and should trigger further investigation.

Bridging therapy prior to axicabtagene ciloleucel for relapsed/refractory large B-cell lymphoma

The impact of bridging therapy (BT) administered between leukapheresis and chimeric antigen receptor (CAR) T-cell therapy for large B-cell lymphoma (LBCL) is unclear. We evaluated the influence of BT (systemic therapy [ST], radiation therapy [RT], or combined-modality therapy [CMT]) on outcomes of 148 LBCL patients who underwent leukapheresis for planned axicabtagene ciloleucel (axi-cel) infusion. The 55% (n = 81) of patients who received BT were more likely to have international prognostic index (IPI) score ≥3 (P ≤ .01), bulky disease (P = .01), and elevated lactate dehydrogenase (LDH; P ≤ .01). The 1-year progression-free (PFS) and overall survival (OS) rates were 40% and 65% in non-BT patients vs 21% and 48% in BT patients (P = .01 and .05, respectively). Twenty-four patients (16%) did not receive axi-cel, most commonly because of lymphoma progression (88%), despite 80% (n = 19) receiving BT. Among 124 patients who received axi-cel, 50% (n = 62) received BT with ST (n = 45), RT (n = 11), or CMT (n = 6); 1-year PFS and OS rates were not significantly different between BT and non-BT cohorts (P = .06 and .21, respectively). There was no difference in proportion of patients with IPI ≥3, limited-stage disease, or elevated LDH between ST, RT, and CMT groups. Compared with non-BT patients, 1-year PFS was inferior for ST-bridged patients (P = .01). RT-bridged patients had improved PFS compared with ST-bridged patients (P = .05). Despite the poor prognosis associated with requiring BT, RT can be an effective bridging strategy. Future studies are necessary to identify strategies that may improve access to CAR T-cell therapy and outcomes.

A systematic review of the quality of conduct and reporting of survival analyses of tuberculosis outcomes in Africa

BACKGROUND

Survival analyses methods (SAMs) are central to analysing time-to-event outcomes. Appropriate application and reporting of such methods are important to ensure correct interpretation of the data. In this study, we systematically review the application and reporting of SAMs in studies of tuberculosis (TB) patients in Africa. It is the first review to assess the application and reporting of SAMs in this context.

METHODS

Systematic review of studies involving TB patients from Africa published between January 2010 and April 2020 in English language. Studies were eligible if they reported use of SAMs. Application and reporting of SAMs were evaluated based on seven author-defined criteria.

RESULTS

Seventy-six studies were included with patient numbers ranging from 56 to 182,890. Forty-three (57%) studies involved a statistician/epidemiologist. The number of published papers per year applying SAMs increased from two in 2010 to 18 in 2019 (P = 0.004). Sample size estimation was not reported by 67 (88%) studies. A total of 22 (29%) studies did not report summary follow-up time. The survival function was commonly presented using Kaplan-Meier survival curves (n = 51, (67%) studies) and group comparisons were performed using log-rank tests (n = 44, (58%) studies). Sixty seven (91%), 3 (4.1%) and 4 (5.4%) studies reported Cox proportional hazard, competing risk and parametric survival regression models, respectively. A total of 37 (49%) studies had hierarchical clustering, of which 28 (76%) did not adjust for the clustering in the analysis. Reporting was adequate among 4.0, 1.3 and 6.6% studies for sample size estimation, plotting of survival curves and test of survival regression underlying assumptions, respectively. Forty-five (59%), 52 (68%) and 73 (96%) studies adequately reported comparison of survival curves, follow-up time and measures of effect, respectively.

CONCLUSION

The quality of reporting survival analyses remains inadequate despite its increasing application. Because similar reporting deficiencies may be common in other diseases in low- and middle-income countries, reporting guidelines, additional training, and more capacity building are needed along with more vigilance by reviewers and journal editors.

Application of Mixture Cure Models in Determining the Survival of Patients with Cervical Cancer

Background: Cervical cancer is the fourth most common cancer among women and the seventh most common worldwide. Objectives: The present study aimed at investigating the usability of cure models in analyzing patients' survival. Besides, the factors affecting the long-term and short-term survival of the patients were determined, using Weibull, log-logistic, and log-normal models. Methods: The sample population of the study included 109 female patients with cervical cancer referred to Motahhari Hospital of Urmia (West Azerbaijan province) from 2004 to 2015. The cure survival analysis was used to determine the patients' survival. Results: The mean and standard deviation age at diagnosis was 50.1 ± 11.7 years. The patients' age, age at marriage, and the disease relapse were significant in the single-variable model on the long-term survival function of the patients. Moreover, the findings showed that Cured Log Logistic Parametric Model was more suitable for analyzing survival data in West Azerbaijan Province, Iran. The relapse variable was significant for all the parametric models. Conclusions: Given the divided sample population into immune and susceptible groups, the mixture cure models can be used to analyze the long-term and short-term survival of the patients with cervical cancer. Moreover, these models can be used to recognize the factors affecting both groups simultaneously.

Explainable machine learning can outperform Cox regression predictions and provide insights in breast cancer survival

Cox Proportional Hazards (CPH) analysis is the standard for survival analysis in oncology. Recently, several machine learning (ML) techniques have been adapted for this task. Although they have shown to yield results at least as good as classical methods, they are often disregarded because of their lack of transparency and little to no explainability, which are key for their adoption in clinical settings. In this paper, we used data from the Netherlands Cancer Registry of 36,658 non-metastatic breast cancer patients to compare the performance of CPH with ML techniques (Random Survival Forests, Survival Support Vector Machines, and Extreme Gradient Boosting [XGB]) in predicting survival using the [Formula: see text]-index. We demonstrated that in our dataset, ML-based models can perform at least as good as the classical CPH regression ([Formula: see text]-index [Formula: see text]), and in the case of XGB even better ([Formula: see text]-index [Formula: see text]). Furthermore, we used Shapley Additive Explanation (SHAP) values to explain the models' predictions. We concluded that the difference in performance can be attributed to XGB's ability to model nonlinearities and complex interactions. We also investigated the impact of specific features on the models' predictions as well as their corresponding insights. Lastly, we showed that explainable ML can generate explicit knowledge of how models make their predictions, which is crucial in increasing the trust and adoption of innovative ML techniques in oncology and healthcare overall.

Patterns of age disparities in colon and lung cancer survival: a systematic narrative literature review

OBJECTIVES

To identify patterns of age disparities in cancer survival, using colon and lung cancer as exemplars.

DESIGN

Systematic review of the literature.

DATA SOURCES

We searched Embase, MEDLINE, Scopus and Web of Science through 18 December 2020.

ELIGIBILITY CRITERIA

We retained all original articles published in English including patients with colon or lung cancer. Eligible studies were required to be population-based, report survival across several age groups (of which at least one was over the age of 65) and at least one other characteristic (eg, sex, treatment).

DATA EXTRACTION AND SYNTHESIS

Two independent reviewers extracted data and assessed the quality of included studies against selected evaluation domains from the QUIPS tool, and items concerning statistical reporting. We evaluated age disparities using the absolute difference in survival or mortality rates between the middle-aged group and the oldest age group, or by describing survival curves.

RESULTS

Out of 3047 references, we retained 59 studies (20 for colon, 34 for lung and 5 for both sites). Regardless of the cancer site, the included studies were highly heterogeneous and often of poor quality. The magnitude of age disparities in survival varied greatly by sex, ethnicity, socioeconomic status, stage at diagnosis, cancer site, and morphology, the number of nodes examined and treatment strategy. Although results were inconsistent for most characteristics, we consistently observed greater age disparities for women with lung cancer compared with men. Also, age disparities increased with more advanced stages for colon cancer and decreased with more advanced stages for lung cancer.

CONCLUSIONS

Although age is one of the most important prognostic factors in cancer survival, age disparities in colon and lung cancer survival have so far been understudied in population-based research. Further studies are needed to better understand age disparities in colon and lung cancer survival.

PROSPERO REGISTRATION NUMBER

CRD42020151402.