Residual mass histology in testicular cancer: development and validation of a clinical prediction rule

Diagnostic Value of the Texture Analysis Parameters of Retroperitoneal Residual Masses on Computed Tomographic Scan after Chemotherapy in Non-Seminomatous Germ Cell Tumors

After chemotherapy, patients with non-seminomatous germ cell tumors (NSGCTs) with residual masses >1 cm on computed tomography (CT) undergo surgery. However, in approximately 50% of cases, these masses only consist of necrosis/fibrosis. We aimed to develop a radiomics score to predict the malignant character of residual masses to avoid surgical overtreatment. Patients with NSGCTs who underwent surgery for residual masses between September 2007 and July 2020 were retrospectively identified from a unicenter database. Residual masses were delineated on post-chemotherapy contrast-enhanced CT scans. Tumor textures were obtained using the free software LifeX. We constructed a radiomics score using a penalized logistic regression model in a training dataset, and evaluated its performance on a test dataset. We included 76 patients, with 149 residual masses; 97 masses were malignant (65%). In the training dataset (n = 99 residual masses), the best model (ELASTIC-NET) led to a radiomics score based on eight texture features. In the test dataset, the area under the curve (AUC), sensibility, and specificity of this model were respectively estimated at 0.82 (95%CI, 0.69-0.95), 90.6% (75.0-98.0), and 61.1% (35.7-82.7). Our radiomics score may help in the prediction of the malignant nature of residual post-chemotherapy masses in NSGCTs before surgery, and thus limit overtreatment. However, these results are insufficient to simply select patients for surgery.

Perioperative complications and oncological outcomes of post-chemotherapy retroperitoneal lymph node dissection in patients with germ cell cancer at two high-volume university centres in Switzerland - a retrospective chart review

BACKGROUND

Post-chemotherapy retroperitoneal lymph node dissection (PC-RPLND) is an integral part of the management of patients with metastatic non-seminoma and residual masses >1 cm after chemotherapy.

AIMS

To assess perioperative complications and oncological outcomes at two major referral centres in Switzerland.

METHODS

This was a retrospective chart review of 136 patients with non-seminoma who underwent PC-RPLND between 2010 and 2020 at the university hospitals of Bern and Zürich. Patient, treatment and tumour characteristics as well as the types and frequencies of intra- and postoperative complications were registered and compared using the chi-square test. Oncological outcomes consisted of the time and location of relapses as well as progression-free and overall survival, which were compared using the log-rank test.

RESULTS

Overall, 70 patients from Bern and 66 patients from Zürich were included; 5 patients had a previous retroperitoneal lymph node dissection (RPLND) (2 Bern, 3 Zürich). Vascular injuries were the most frequent intraoperative complication, occurring in 27/136 (19.9%) patients. Postoperative complications were observed in 42/136 (30.9%) patients, ileus being the most common. Perioperative mortality was 2.2%. A retroperitoneal mass ≥50 mm was significantly associated with intraoperative complications (p = 0.004) and increased resource demands (p = 0.021). Postoperative morbidity was higher according to age at post-chemotherapy retroperitoneal lymph node dissection ≥40 years (p = 0.028) and retroperitoneal mass ≥20 mm (p = 0.005). The median follow-up time was 37 months (interquartile range [IQR] 18-64 months). The median progression-free survival at 5 years was 76% (95% confidence interval [CI]: 64-85%) in Bern and 69% (95% CI: 54-80%) in Zürich (p = 0.464). The median overall survival at 5 years was 88% (95% CI: 76-94%) in Bern and 77% (95% CI: 60-87%) in Zürich (p = 0.335). Patients with progressive disease or a tumour marker increase before retroperitoneal lymph node dissection had significantly inferior progression-free and overall survival compared to non-progressing patients. The presence of teratoma in resected specimens did not confer inferior survival probabilities compared to necrosis only, whereas the presence of vital undifferentiated tumour conferred inferior progression-free and overall survival. Patients with a previous retroperitoneal lymph node dissection and patients operated for late relapses >2 years after chemotherapy also had significantly inferior progression-free and overall survival.

CONCLUSIONS

We found a relevant rate of severe perioperative complications at PC-RPLND at even experienced high-volume centres. The oncological outcomes at two major university urological centres in Switzerland were similar and determined by preoperative risk factors and intraoperative histology.

Prediction model for 3-year rupture risk of unruptured cerebral aneurysms in Japanese patients

OBJECTIVE

To build a prediction model that estimates the 3-year rupture risk of unruptured saccular cerebral aneurysms.

METHODS

Survival analysis was done using each aneurysm as the unit for analysis. Derivation data were from the Unruptured Cerebral Aneurysm Study (UCAS) in Japan. It consists of patients with unruptured cerebral aneurysms enrolled between 2000 and 2004 at neurosurgical departments at tertiary care hospitals in Japan. The model was presented as a scoring system, and aneurysms were classified into 4 risk grades by predicted 3-year rupture risk: I, < 1%; II, 1 to 3%; III, 3 to 9%, and IV, >9%. The discrimination property and calibration plot of the model were evaluated with external validation data. They were a combination of 3 Japanese cohort studies: UCAS II, the Small Unruptured Intracranial Aneurysm Verification study, and the study at Jikei University School of Medicine.

RESULTS

The derivation data include 6,606 unruptured cerebral aneurysms in 5,651 patients. During the 11,482 aneurysm-year follow-up period, 107 ruptures were observed. The predictors chosen for the scoring system were patient age, sex, and hypertension, along with aneurysm size, location, and the presence of a daughter sac. The 3-year risk of rupture ranged from <1% to >15% depending on the individual characteristics of patients and aneurysms. External validation indicated good discrimination and calibration properties.

INTERPRETATION

A simple scoring system that only needs easily available patient and aneurysmal information was constructed. This can be used in clinical decision making regarding management of unruptured cerebral aneurysms.

Reporting and methods in clinical prediction research: a systematic review

BACKGROUND

We investigated the reporting and methods of prediction studies, focusing on aims, designs, participant selection, outcomes, predictors, statistical power, statistical methods, and predictive performance measures.

METHODS AND FINDINGS

We used a full hand search to identify all prediction studies published in 2008 in six high impact general medical journals. We developed a comprehensive item list to systematically score conduct and reporting of the studies, based on recent recommendations for prediction research. Two reviewers independently scored the studies. We retrieved 71 papers for full text review: 51 were predictor finding studies, 14 were prediction model development studies, three addressed an external validation of a previously developed model, and three reported on a model's impact on participant outcome. Study design was unclear in 15% of studies, and a prospective cohort was used in most studies (60%). Descriptions of the participants and definitions of predictor and outcome were generally good. Despite many recommendations against doing so, continuous predictors were often dichotomized (32% of studies). The number of events per predictor as a measure of statistical power could not be determined in 67% of the studies; of the remainder, 53% had fewer than the commonly recommended value of ten events per predictor. Methods for a priori selection of candidate predictors were described in most studies (68%). A substantial number of studies relied on a p-value cut-off of p<0.05 to select predictors in the multivariable analyses (29%). Predictive model performance measures, i.e., calibration and discrimination, were reported in 12% and 27% of studies, respectively.

CONCLUSIONS

The majority of prediction studies in high impact journals do not follow current methodological recommendations, limiting their reliability and applicability.

Assessing the performance of prediction models: a framework for traditional and novel measures

The performance of prediction models can be assessed using a variety of methods and metrics. Traditional measures for binary and survival outcomes include the Brier score to indicate overall model performance, the concordance (or c) statistic for discriminative ability (or area under the receiver operating characteristic [ROC] curve), and goodness-of-fit statistics for calibration.Several new measures have recently been proposed that can be seen as refinements of discrimination measures, including variants of the c statistic for survival, reclassification tables, net reclassification improvement (NRI), and integrated discrimination improvement (IDI). Moreover, decision-analytic measures have been proposed, including decision curves to plot the net benefit achieved by making decisions based on model predictions.We aimed to define the role of these relatively novel approaches in the evaluation of the performance of prediction models. For illustration, we present a case study of predicting the presence of residual tumor versus benign tissue in patients with testicular cancer (n = 544 for model development, n = 273 for external validation).We suggest that reporting discrimination and calibration will always be important for a prediction model. Decision-analytic measures should be reported if the predictive model is to be used for clinical decisions. Other measures of performance may be warranted in specific applications, such as reclassification metrics to gain insight into the value of adding a novel predictor to an established model.

Retroperitoneal lymph node dissection after chemotherapy

Retroperitoneal lymph node dissection after chemotherapy (PC-RPLND) plays a crucial role in managing patients with advanced germ cell tumours (GCTs). In the last few years improvements in radiographic staging, a better understanding of the role of serum tumour markers, and the introduction of cisplatin-based chemotherapy have all contributed to this surgical therapy. PC-RPLND is necessary when residual radiographic abnormalities are present after chemotherapy. The need for a PC-RPLND in the face of normal findings from computed tomography (CT) is controversial. CT criteria alone are not sufficiently reliable to distinguish viable tumour or teratoma from necrosis. No combination of variables can predict negative retroperitoneal pathology with sufficient accuracy after induction chemotherapy. Unresected teratoma or viable GCT are at least partly chemorefractory and, if untreated, will progress. So completeness of resection is an independent and consistent predictive variable of clinical outcome. In PC-RPLND surgical margins should not be compromised in an attempt to preserve ejaculation, although nerve-sparing dissections are possible in patients with marker normalization after chemotherapy and necrotic tissue in frozen-section histology. In these patients nerve-sparing techniques and the reduction of surgical field to the left- or right-sided template are applicable to preserve antegrade ejaculation and consecutive fertility. The size and location of residual masses coupled with the retroperitoneal desmoplastic reaction make PC-RPLND a technically demanding procedure that should be performed by experienced surgeons in dedicated referral centres.

Profile-specific survival estimates: Making reports of clinical trials more patient-relevant

Background When considering treatment options, a physician needs to know the prognosis corresponding to the risk profile of the patient seeking treatment. Reports of clinical trials generally address treatment-specific survival probabilities only in the aggregate, i.e., for the typical patient, and often express the difference in survival as a hazard ratio. Such summaries do not provide treatment-specific survival probabilities (and thus the absolute difference in these probabilities) for patient profiles that are not near the typical of those in the trial. Despite the fact that Cox intended his hazard regression method to be used to produce such profile-specific survival estimates, and even showed how to calculate them, authors are either unaware that this is possible, or else choose not to report them.

Purpose To illustrate how treatment- and profile-specific survival estimates are obtained from the Cox method, and can be displayed in a compact form.

Methods We derive treatment- and profile-specific survival probabilities from the estimated survival function for the `reference' profile. Data from the Systolic Hypertension in the Elderly Program study serve as an illustration.

Results Two different formats, tabular and nomogram-based, allow the entire set of estimated treatment- and profile-specific survival probabilities to be reported.

Limitations Estimates are limited to the profiles within the covariate-space spanned by the trial, and depend on the correctness of the model.

Conclusion Treatment- and profile-specific survival estimates are practice-relevant, almost never reported, estimable from the Cox model, and easy to report in a compact form. Clinical Trials 2008; 5: 107—115. http://ctj.sagepub.com

Polytomous logistic regression analysis could be applied more often in diagnostic research

OBJECTIVE

Physicians commonly consider the presence of all differential diagnoses simultaneously. Polytomous logistic regression modeling allows for simultaneous estimation of the probability of multiple diagnoses. We discuss and (empirically) illustrate the value of this method for diagnostic research.

STUDY DESIGN AND SETTING

We used data from a study on the diagnosis of residual retroperitoneal mass histology in patients presenting with nonseminomatous testicular germ cell tumor. The differential diagnoses include benign tissue, mature teratoma, and viable cancer. Probabilities of each diagnosis were estimated with a polytomous logistic regression model and compared with the probabilities estimated from two consecutive dichotomous logistic regression models.

RESULTS

We provide interpretations of the odds ratios derived from the polytomous regression model and present a simple score chart to facilitate calculation of predicted probabilities from the polytomous model. For both modeling methods, we show the calibration plots and receiver operating characteristics curve (ROC) areas comparing each diagnostic outcome category with the other two. The ROC areas for benign tissue, mature teratoma, and viable cancer were similar for both modeling methods, 0.83 (95% confidence interval [CI]=0.80-0.85) vs. 0.83 (95% CI=0.80-0.85), 0.78 (95% CI=0.75-0.81) vs. 0.78 (95% CI=0.75-0.81), and 0.66 (95% CI=0.61-0.71) vs. 0.64 (95% CI=0.59-0.69), for polytomous and dichotomous regression models, respectively.

CONCLUSION

Polytomous logistic regression is a useful technique to simultaneously model predicted probabilities of multiple diagnostic outcome categories. The performance of a polytomous prediction model can be assessed similarly to a dichotomous logistic regression model, and predictions by a polytomous model can be made with a user-friendly method. Because the simultaneous consideration of the presence of multiple (differential) conditions serves clinical practice better than consideration of the presence of only one target condition, polytomous logistic regression could be applied more often in diagnostic research.

Metastatic lymph nodes in urogenital cancers: contribution of imaging findings

The presence of nodal metastasis in patients with urogenital malignancies is an important factor for prognosis, and radiologic identification of enlarged nodes greatly affects treatment choices. Radiologic evaluation for nodal metastases is usually performed with computed tomography, but magnetic resonance imaging is also useful in evaluating primary and nodal metastases in pelvic tumors. On these cross-sectional modalities, nodal metastases are usually suspected according to location and size criteria. Although there has been no general consensus on the criteria, a short axis diameter of 8 to 10 mm is generally applied. However, radiologic evaluation does not always provide sufficient accuracy for nodal staging because of an inability to diagnose smaller metastatic lymph nodes. The clinical significance of a radiologic recognition of enlarged nodes also differs by cancer type in relation to differences in staging systems and treatment. The presence of regional lymphadenopathy in patients with renal cell carcinoma often alters surgical methods, whereas the presence of regional lymphadenopathy is an indication of systemic chemotherapy in patients with cancers of the urinary tract, prostate, and testicles. In this report, preferential sites and staging of nodal metastasis and contributions of radiologic imaging are reviewed for each urogenital cancer.

Substantial effective sample sizes were required for external validation studies of predictive logistic regression models

BACKGROUND AND OBJECTIVES

The performance of a prediction model is usually worse in external validation data compared to the development data. We aimed to determine at which effective sample sizes (i.e., number of events) relevant differences in model performance can be detected with adequate power.

METHODS

We used a logistic regression model to predict the probability that residual masses of patients treated for metastatic testicular cancer contained only benign tissue. We performed standard power calculations and Monte Carlo simulations to estimate the numbers of events that are required to detect several types of model invalidity with 80% power at the 5% significance level.

RESULTS

A validation sample with 111 events was required to detect that a model predicted too high probabilities, when predictions were on average 1.5 times too high on the odds scale. A decrease in discriminative ability of the model, indicated by a decrease in the c-statistic from 0.83 to 0.73, required 81 to 106 events, depending on the specific scenario.

CONCLUSION

We suggest a minimum of 100 events and 100 nonevents for external validation samples. Specific hypotheses may, however, require substantially higher effective sample sizes to obtain adequate power.