Are we able to predict survival in ER-positive HER2-negative breast cancer? A comparison of web-based models

Machine learning survival models trained on clinical data to identify high risk patients with hormone responsive HER2 negative breast cancer

For endocrine-positive Her2 negative breast cancer patients at an early stage, the benefit of adding chemotherapy to adjuvant endocrine therapy is not still confirmed. Several genomic tests are available on the market but are very expensive. Therefore, there is the urgent need to explore novel reliable and less expensive prognostic tools in this setting. In this paper, we shown a machine learning survival model to estimate Invasive Disease-Free Events trained on clinical and histological data commonly collected in clinical practice. We collected clinical and cytohistological outcomes of 145 patients referred to Istituto Tumori "Giovanni Paolo II". Three machine learning survival models are compared with the Cox proportional hazards regression according to time-dependent performance metrics evaluated in cross-validation. The c-index at 10 years obtained by random survival forest, gradient boosting, and component-wise gradient boosting is stabled with or without feature selection at approximately 0.68 in average respect to 0.57 obtained to Cox model. Moreover, machine learning survival models have accurately discriminated low- and high-risk patients, and so a large group which can be spared additional chemotherapy to hormone therapy. The preliminary results obtained by including only clinical determinants are encouraging. The integrated use of data already collected in clinical practice for routine diagnostic investigations, if properly analyzed, can reduce time and costs of the genomic tests.

The IHC4+C score: an affordable and reproducible non-molecular decision-aid in hormone receptor-positive breast cancer. Does it still hold value for patients in 2020?

AIM

The majority of women diagnosed with early breast cancer have hormone-receptor positive (HR+)/HER2-negative disease. Adjuvant endocrine therapy provides substantial risk reduction benefits in virtually all patients. The role of adjuvant chemotherapy in certain subsets of patients is equivocal. This paper sought to evaluate the role of the IHC4+C score to aid this clinical decision in addition to providing an overview of the current molecular and non- molecular tools available in the adjuvant setting.

METHODS

This prospective study included 53 post-operative HR+/HER2- negative early breast cancer patients selected from the multidiscipliniary team meeting between August 2017 and January 2020. IHC4+C testing was requested by clinicians for patients in whom the availability of the score may have impacted adjuvant decision-making. Adjuvant treatment decisions were recorded at three time points (prior and post IHC4+C scoring as well as the patient's final decision). The primary goal was the proportion of patients who were spared chemotherapy following the availability of IHC4+C scores to impact on clinicians' recommendations for adjuvant systemic therapy.

RESULTS

A total of 34 patients (64%) were initially recommended to undergo chemotherapy or to consider chemotherapy. With the availability of the IHC4+C score, only 17 patients (32%) underwent chemotherapy, demonstrating a substantial reduction in the frequency of chemotherapy prescribing.

CONCLUSION

This study demonstrates that when utilized appropriately in a multidisciplinary setting, the IHC4+C algorithm is an alternative, reproducible and affordable tool with a proven capacity to stratify risk and to spare a large proportion of patients from undergoing chemotherapy.

Determination of breast cancer prognosis after neoadjuvant chemotherapy: comparison of Residual Cancer Burden (RCB) and Neo-Bioscore

Background

To compare RCB (Residual Cancer Burden) and Neo-Bioscore in terms of prognostic performance and see if adding pathological variables improve these scores.

Methods

We analysed 750 female patients with invasive breast cancer (BC) treated with neoadjuvant chemotherapy (NAC) at Institut Curie between 2002 and 2012. Scores were compared in global population and by BC subtype using Akaike information criterion (AIC), C-Index (concordance index), calibration curves and after adding lymphovascular invasion (LVI) and pre-/post-NAC TILs levels.

Results

RCB and Neo-Bioscore were significantly associated to disease-free and overall survival in global population and for triple-negative BC. RCB had the lowest AICs in every BC subtype, corresponding to a better prognostic performance. In global population, C-Index values were poor for RCB (0.66; CI [0.61–0.71]) and fair for Neo-Bioscore (0.70; CI [0.65–0.75]). Scores were well calibrated in global population, but RCB yielded better prognostic performances in each BC subtype. Concordance between the two scores was poor. Adding LVI and TILs improved the performance of both scores.

Conclusions

Although RCB and Neo-Bioscore had similar prognostic performances, RCB showed better performance in BC subtypes, especially in luminal and TNBC. By generating fewer prognostic categories, RCB enables an easier use in everyday clinical practice.

A Comparison Between the Online Prediction Models CancerMath and PREDICT as Prognostic Tools in Thai Breast Cancer Patients

Background and Purpose

Web-based prognostic calculators have been developed to inform about the use of adjuvant systemic treatments in breast cancer. CancerMath and PREDICT are two examples of web-based prognostic tools that predict patient survival up to 15 years after an initial diagnosis of breast cancer. The aim of this study is to validate the use of CancerMath and PREDICT as prognostic tools in Thai breast cancer patients.

Patients and Methods

A total of 615 patients who underwent surgical treatment for stage I to III breast cancer from 2003 to 2011 at the Division of Head Neck and Breast Surgery, Department of Surgery, Siriraj Hospital, Mahidol University, Thailand were recruited. A model-predicted overall survival rate (OS) and the actual OS of the patients were compared. The efficacy of the model was evaluated using receiver-operating characteristic (ROC) analysis.

Results

For CancerMath, the predicted 5-year OS was 88.9% and the predicted 10-year OS was 78.3% (p<0.001). For PREDICT, the predicted 5-year OS was 83.1% and the predicted 10-year OS was 72.0% (p<0.001). The actual observed 5-year OS was 90.8% and the observed 10-year OS was 82.6% (p<0.001). CancerMath demonstrated better predictive performance than PREDICT in all subgroups for both 5- and 10-year OS. In addition, there was a marked difference between CancerMath and observed survival rates in patients who were older as well as patients who were stage N3. The area under the ROC curve for 5-year OS in CancerMath and 10-year OS was 0.74 (95% CI; 0.65-0.82) and 0.75 (95% CI; 0.68-0.82). In the PREDICT group, the area under the ROC curve for 5-year OS was 0.78 (95% CI; 0.71-0.85) and for 10-year OS, it was 0.78 (95% CI; 0.71-0.84).

Conclusion

CancerMath and PREDICT models both underestimated the OS in Thai breast cancer patients. Thus, a novel prognostic model for Thai breast cancer patients is required.

Validation of the online prediction model CancerMath in the Dutch breast cancer population

PURPOSE

CancerMath predicts the expected benefit of adjuvant systemic therapy on overall (OS) and breast cancer-specific survival (BCSS). Here, CancerMath was validated in Dutch breast cancer patients.

METHODS

All operated women diagnosed with stage I-III primary invasive breast cancer in 2005 were identified from the Netherlands Cancer Registry. Calibration was assessed by comparing 5- and 10-year predicted and observed OS/BCSS using χ² tests. A difference > 3% was considered as clinically relevant. Discrimination was assessed by area under the receiver operating characteristic (AUC) curves.

RESULTS

Altogether, 8032 women were included. CancerMath underestimated 5- and 10-year OS by 2.2% and 1.9%, respectively. AUCs of 5- and 10-year OS were both 0.77. Divergence between predicted and observed OS was most pronounced in grade II, patients without positive nodes, tumours 1.01-2.00 cm, hormonal receptor positive disease and patients 60-69 years. CancerMath underestimated 5- and 10-year BCSS by 0.5% and 0.6%, respectively. AUCs were 0.78 and 0.73, respectively. No significant difference was found in any subgroup.

CONCLUSION

CancerMath predicts OS accurately for most patients with early breast cancer although outcomes should be interpreted with care in some subgroups. BCSS is predicted accurately in all subgroups. Therefore, CancerMath can reliably be used in (Dutch) clinical practice.

The long noncoding RNA MIR210HG promotes tumor metastasis by acting as a ceRNA of miR-1226-3p to regulate mucin-1c expression in invasive breast cancer

Background: Long noncoding RNAs have been known to be involved in multiple types of malignancies, including invasive breast cancer (IBC). This study aimed to explore the role of long noncoding RNAs in IBC and elucidate the potential molecular mechanisms.

Methods: Using TCGA microarray data analysis, we identified a long noncoding RNA, MIR210HG, highly expressed in IBC. Kaplan-Meier method and the log-rank test were used for survival analysis. The gain-of-function experiments were performed to assess the function of MIR210HG in IBC invasion and migration in both in vitro and in vivo settings. Bioinformatic analysis as well as luciferase reporter assay, rescue experiments and western blot assay revealed the mode of action of MIR210HG.

Results: The aberrantly enhanced MiR210HG expression predicted poor prognosis and lower survival rate. Knockdown of MiR210HG suppressed IBC cell invasion and metastasis both in vitro and in vivo. MiR-1226-3p was identified and validated to be the target miRNA of MiR210HG. Furthermore, MiR210HG functions as a competing endogenous RNAs (ceRNA) which sponges miR-1226-3p, therefore upregulates the expression of mucin1 (MUC1-C).

Conclusions: Our study demonstrated that MiR210HG sponges miR-1226-3p to facilitate invasive breast cancer cell invasion and metastasis by regulating mucin-1c and EMT pathway, revealing the oncogenic role of MiR210HG in IBC cells.

Communicating prognosis to women with early breast cancer - overview of prediction tools and the development and pilot testing of a decision aid

Background

Shared decision-making in oncology requires information on individual prognosis. This comprises cancer prognosis as well as competing risks of dying due to age and comorbidities. Decision aids usually do not provide such information on competing risks. We conducted an overview on clinical prediction tools for early breast cancer and developed and pilot-tested a decision aid (DA) addressing individual prognosis using additional chemotherapy in early, hormone receptor-positive breast cancer as an example.

Methods

Systematic literature search on clinical prediction tools for the effects of drug treatment on survival of breast cancer. The DA was developed following criteria for evidence-based patient information and International Patient Decision Aids Standards. We included data on the influence of age and comorbidities on overall prognosis. The DA was pilot-tested in focus groups. Comprehension was additionally evaluated through an online survey with women in breast cancer self-help groups.

Results

We identified three prediction tools: Adjuvant!Online, PREDICT and CancerMath. All tools consider age and tumor characteristics. Adjuvant!Online considers comorbidities, CancerMath displays age-dependent non-cancer mortality. Harm due to therapy is not reported.

Twenty women participated in focus groups piloting the DA until data saturation was achieved. A total of 102 women consented to participate in the online survey, of which 86 completed the survey. The rate of correct responses was 90.5% and ranged between 84 and 95% for individual questions.

Conclusions

None of the clinical prediction tools fulfilled the requirements to provide women with all the necessary information for informed decision-making. Information on individual prognosis was well understood and can be included in patient decision aids.

Electronic supplementary material

The online version of this article (10.1186/s12913-019-3988-2) contains supplementary material, which is available to authorized users.

Prognostic models for breast cancer: a systematic review

BACKGROUND

Breast cancer is the most common cancer in women worldwide, with a great diversity in outcomes among individual patients. The ability to accurately predict a breast cancer outcome is important to patients, physicians, researchers, and policy makers. Many models have been developed and tested in different settings. We systematically reviewed the prognostic models developed and/or validated for patients with breast cancer.

METHODS

We conducted a systematic search in four electronic databases and some oncology websites, and a manual search in the bibliographies of the included studies. We identified original studies that were published prior to 1st January 2017, and presented the development and/or validation of models based mainly on clinico-pathological factors to predict mortality and/or recurrence in female breast cancer patients.

RESULTS

From the 96 articles selected from 4095 citations found, we identified 58 models, which predicted mortality (n = 28), recurrence (n = 23), or both (n = 7). The most frequently used predictors were nodal status (n = 49), tumour size (n = 42), tumour grade (n = 29), age at diagnosis (n = 24), and oestrogen receptor status (n = 21). Models were developed in Europe (n = 25), Asia (n = 13), North America (n = 12), and Australia (n = 1) between 1982 and 2016. Models were validated in the development cohorts (n = 43) and/or independent populations (n = 17), by comparing the predicted outcomes with the observed outcomes (n = 55) and/or with the outcomes estimated by other models (n = 32), or the outcomes estimated by individual prognostic factors (n = 8). The most commonly used methods were: Cox proportional hazards regression for model development (n = 32); the absolute differences between the predicted and observed outcomes (n = 30) for calibration; and C-index/AUC (n = 44) for discrimination. Overall, the models performed well in the development cohorts but less accurately in some independent populations, particularly in patients with high risk and young and elderly patients. An exception is the Nottingham Prognostic Index, which retains its predicting ability in most independent populations.

CONCLUSIONS

Many prognostic models have been developed for breast cancer, but only a few have been validated widely in different settings. Importantly, their performance was suboptimal in independent populations, particularly in patients with high risk and in young and elderly patients.

Development and validation of a new predictive model for breast cancer survival in New Zealand and comparison to the Nottingham prognostic index

Background

The only available predictive models for the outcome of breast cancer patients in New Zealand (NZ) are based on data in other countries. We aimed to develop and validate a predictive model using NZ data for this population, and compare its performance to a widely used overseas model, the Nottingham Prognostic Index (NPI).

Methods

We developed a model to predict 10-year breast cancer-specific survival, using data collected prospectively in the largest population-based regional breast cancer registry in NZ (Auckland, 9182 patients), and assessed its performance in this data set (internal validation) and in an independent NZ population-based series of 2625 patients in Waikato (external validation). The data included all women with primary invasive breast cancer diagnosed from 1 June 2000 to 30 June 2014, with follow up to death or Dec 31, 2014. We used multivariate Cox proportional hazards regression to assess predictors and to calculate predicted 10-year breast cancer mortality, and therefore survival, probability for each patient. We assessed observed survival by the Kaplan Meier method. We assessed discrimination by the C statistic, and calibration by comparing predicted and observed survival rates for patients in 10 groups ordered by predicted 10-year survival. We compared this NZ model with the Nottingham Prognostic Index (NPI) in this validation data set.

Results

Discrimination was good: C statistics were 0.84 for internal validity and 0.83 for an independent external validity. For calibration, for both internal and external validity the predicted 10-year survival probabilities in all groups of patients, ordered by predicted survival, were within the 95% confidence intervals (CI) of the observed Kaplan-Meier survival probabilities. The NZ model showed good discrimination even within the prognostic groups defined by the NPI.

Conclusions

These results for the New Zealand model show good internal and external validity, transportability, and potential clinical value of the model, and its clear superiority over the NPI. Further research is needed to assess other potential predictors, to assess the model’s performance in specific subgroups of patients, and to compare it to other models, which have been developed in other countries and have not yet been tested in NZ.

Electronic supplementary material

The online version of this article (10.1186/s12885-018-4791-x) contains supplementary material, which is available to authorized users.

Development and External Validation of Prediction Models for 10-Year Survival of Invasive Breast Cancer. Comparison with PREDICT and CancerMath

Purpose: To compare PREDICT and CancerMath, two widely used prognostic models for invasive breast cancer, taking into account their clinical utility. Furthermore, it is unclear whether these models could be improved.Experimental Design: A dataset of 5,729 women was used for model development. A Bayesian variable selection algorithm was implemented to stochastically search for important interaction terms among the predictors. The derived models were then compared in three independent datasets (n = 5,534). We examined calibration, discrimination, and performed decision curve analysis.Results: CancerMath demonstrated worse calibration performance compared with PREDICT in estrogen receptor (ER)-positive and ER-negative tumors. The decline in discrimination performance was -4.27% (-6.39 to -2.03) and -3.21% (-5.9 to -0.48) for ER-positive and ER-negative tumors, respectively. Our new models matched the performance of PREDICT in terms of calibration and discrimination, but offered no improvement. Decision curve analysis showed predictions for all models were clinically useful for treatment decisions made at risk thresholds between 5% and 55% for ER-positive tumors and at thresholds of 15% to 60% for ER-negative tumors. Within these threshold ranges, CancerMath provided the lowest clinical utility among all the models.Conclusions: Survival probabilities from PREDICT offer both improved accuracy and discrimination over CancerMath. Using PREDICT to make treatment decisions offers greater clinical utility than CancerMath over a range of risk thresholds. Our new models performed as well as PREDICT, but no better, suggesting that, in this setting, including further interaction terms offers no predictive benefit. Clin Cancer Res; 24(9); 2110-5. ©2018 AACR.