Correlative studies of the Breast Cancer Index (HOXB13/IL17BR) and ER, PR, AR, AR/ER ratio and Ki67 for prediction of extended endocrine therapy benefit: a Trans-aTTom study.

BACKGROUND: Multiple clinical trials demonstrate consistent but modest benefit of adjuvant extended endocrine therapy (EET) in HR + breast cancer patients. Predictive biomarkers to identify patients that benefit from EET are critical to balance modest reductions in risk against potential side effects of EET. This study compares the performance of the Breast Cancer Index, BCI (HOXB13/IL17BR, H/I), with expression of estrogen (ER), progesterone (PR), and androgen receptors (AR), and Ki67, for prediction of EET benefit. METHODS: Node-positive (N+) patients from the Trans-aTTom study with available tissue specimen and BCI results (N = 789) were included. Expression of ER, PR, AR, and Ki67 was assessed by quantitative immunohistochemistry. BCI (H/I) gene expression analysis was conducted by quantitative RT-PCR. Statistical significance of the treatment by biomarker interaction was evaluated by likelihood ratio tests based on multivariate Cox proportional models, adjusting for age, tumor size, grade, and HER2 status. Pearson's correlation coefficients were calculated to evaluate correlations between BCI (H/I) versus ER, PR, AR, Ki67 and AR/ER ratio. RESULTS: EET benefit, measured by the difference in risk of recurrence between patients treated with tamoxifen for 10 versus 5 years, is significantly associated with increasing values of BCI (H/I) (interaction P = 0.01). In contrast, expression of ER (P = 0.83), PR (P = 0.66), AR (P = 0.78), Ki67 (P = 0.87) and AR/ER ratio (P = 0.84) exhibited no significant relationship with EET benefit. BCI (H/I) showed a very weak negative correlation with ER (r = - 0.18), PR (r = - 0.25), and AR (r = - 0.14) expression, but no correlation with either Ki67 (r = 0.04) or AR/ER ratio (r = 0.02). CONCLUSION: These findings are consistent with the growing body of evidence that BCI (H/I) is significantly predictive of response to EET and outcome. Results from this direct comparison demonstrate that expression of ER, PR, AR, Ki67 or AR/ER ratio are not predictive of benefit from EET. BCI (H/I) is the only clinically validated biomarker that predicts EET benefit.

Breast Cancer Index Is a Predictive Biomarker of Treatment Benefit and Outcome from Extended Tamoxifen Therapy: Final Analysis of the Trans-aTTom Study.

PURPOSE: The Breast Cancer Index (BCI) HOXB13/IL17BR (H/I) ratio predicts benefit from extended endocrine therapy in hormone receptor-positive (HR+) early-stage breast cancer. Here, we report the final analysis of the Trans-aTTom study examining BCI (H/I)'s predictive performance. EXPERIMENTAL DESIGN: BCI results were available for 2,445 aTTom trial patients. The primary endpoint of recurrence-free interval (RFI) and secondary endpoints of disease-free interval (DFI) and disease-free survival (DFS) were examined using Cox proportional hazards regression and log-rank test. RESULTS: Final analysis of the overall study population (N = 2,445) did not show a significant improvement in RFI with extended tamoxifen [HR, 0.90; 95% confidence interval (CI), 0.69-1.16; P = 0.401]. Both the overall study population and N0 group were underpowered due to the low event rate in the N0 group. In a pre-planned analysis of the N+ subset (N = 789), BCI (H/I)-High patients derived significant benefit from extended tamoxifen (9.7% absolute benefit: HR, 0.33; 95% CI, 0.14-0.75; P = 0.016), whereas BCI (H/I)-Low patients did not (-1.2% absolute benefit; HR, 1.11; 95% CI, 0.76-1.64; P = 0.581). A significant treatment-to-biomarker interaction was demonstrated on the basis of RFI, DFI, and DFS (P = 0.037, 0.040, and 0.025, respectively). BCI (H/I)-High patients remained predictive of benefit from extended tamoxifen in the N+/HER2- subgroup (9.4% absolute benefit: HR, 0.35; 95% CI, 0.15-0.81; P = 0.047). A three-way interaction evaluating BCI (H/I), treatment, and HER2 status was not statistically significant (P = 0.849). CONCLUSIONS: Novel findings demonstrate that BCI (H/I) significantly predicts benefit from extended tamoxifen in HR+ N+ patients with HER2- disease. Moreover, BCI (H/I) demonstrates significant treatment to biomarker interaction across survival outcomes.

Topical treatment of actinic keratoses in organ transplant recipients: a feasibility study for SPOT (Squamous cell carcinoma Prevention in Organ transplant recipients using Topical treatments).

BACKGROUND: The risk of cutaneous squamous cell carcinoma (cSCC) is significantly increased in organ transplant recipients (OTRs). Clearance of actinic keratoses (AKs) is generally regarded as a surrogate biomarker for cSCC prevention. OTR-cSCC chemoprevention with topical AK treatments has not been investigated in randomized controlled trials (RCTs), although there is evidence that 5% 5-fluorouracil (5-FU) may be chemoprotective in immunocompetent patients. OBJECTIVES: To assess the feasibility, activity and evaluation outcomes relevant to the design of a future phase III RCT of topical cSCC chemoprevention in OTRs. METHODS: OTRs with 10 or more AKs in predefined areas were randomized 1 : 1 : 1 to topical 5-FU, 5% imiquimod (IMIQ) or sunscreen (sun-protective factor 30+) in a phase II, open-label RCT over 15 months. Feasibility outcomes included proportions of eligible OTRs randomized, completing treatment and willing to be re-treated. AK activity [AK clearance, new AK development, patient-centred outcomes (toxicity, health-related quality of life, HRQoL)] and evaluation methodology (clinical vs. photographic) were assessed. RESULTS: Forty OTRs with 903 AKs were randomized. All feasibility outcomes were met (56% of eligible OTRs were randomized; 89% completed treatment; 81% were willing to be re-treated). AK activity analyses found 5-FU and IMIQ were superior to sunscreen for AK clearance and prevention of new AKs. 5-FU was more effective than IMIQ in AK clearance and prevention in exploratory analyses. Although toxicity was greater with 5-FU, HRQoL outcomes were similar. CONCLUSIONS: Trials of topical AK treatments in OTRs for cSCC chemoprevention are feasible and AK activity results support further investigation of 5-FU-based treatments in future phase III trials. What is already known about this topic? Cutaneous squamous cell carcinoma (cSCC) is significantly more common in immunocompromised individuals including organ transplant recipients (OTRs) compared with immunocompetent populations. cSCC chemoprevention activity of sunscreen and 5-fluorouracil-based (5-FU) actinic keratosis (AK) treatments has been demonstrated in randomized controlled trials (RCTs) in immunocompetent populations but not in OTRs. AKs are cSCC precursors and their clearance and prevention are generally regarded as surrogate endpoint biomarkers for potential cSCC chemoprevention activity. What does this study add? SPOT (SCC Prevention in OTRs using Topical treatments) has confirmed that RCTs of OTR-cSCC chemoprevention with topical AK treatments are feasible. It also suggests that topical 5-FU may be superior to 5% imiquimod and sunscreen in AK clearance and prevention. Together with recent evidence from several RCTs in the general population, these data provide a compelling rationale for further studies of intervention with 5-FU-based topical chemoprevention approaches in OTR-cSCC prevention.

Evidence for the use of spinal collars in stabilising spinal injuries in the pre-hospital setting in trauma patients: a systematic review.

PURPOSE: Spinal collars were introduced in 1967 into the management of spinal trauma care as it was thought that this technique of immobilisation would prevent any further neurological or spinal damage in high-risk patients. The aim of this systematic review was to determine whether the use of spinal collars in the pre-hospital trauma patient was recommended by published literature. METHODS: A systematic search of the literature was conducted between 1990 and 2020, screening PubMed, Medline, Science Direct and Google Scholar. The consequent findings were then qualitatively synthesised with the aim of effectively evaluating the evidence to resolve the discrepancy between current practice and literature. RESULTS: Of the nine eligible studies, six deemed that spinal collars should not be used in pre-hospital trauma patients with the remaining three reporting uncertainty if spinal collars were best practice. Our results suggest that there is a discrepancy between current guidance and practice in that although the guidelines recommend the use of spinal collars in the pre-hospital setting the majority of the studies were against the use of spinal collars. Importantly, none of the studies reported any benefits of spinal collars. CONCLUSION: Our study shows a disparity between current guidelines and the published literature and warrants further direct research to obtain a more comprehensive view of the use of spinal collars in a pre-hospital setting.

Azacitidine for the treatment of steroid-refractory chronic graft-versus-host disease: the results of the phase II AZTEC clinical trial.

Chronic graft-versus-host disease (cGvHD) is a major cause of non-relapse morbidity and mortality following allogeneic stem cell transplant. Over half of patients with moderate or severe cGvHD fail to respond adequately to first-line treatment with systemic steroids, and although a range of second-line options have been employed, a lack of prospective evidence means there is no standard of care. The AZTEC trial is a prospective, single-arm, phase II study investigating the safety and activity of azacitidine for the treatment of cGvHD in patients who are resistant to, or intolerant of, systemic steroid therapy. The co-primary outcomes were treatment tolerability, and activity measured as objective response according to modified National Institutes of Health criteria. Fourteen patients were recruited to the first stage of the trial, of whom seven completed the planned six cycles of azacitidine 36 mg/m2 days 1-5 per 28-day cycle. Azacitidine was tolerated by 13/14 patients, and 7/14 showed an objective response. Clinical responses were mirrored by improvements in patient-reported cGvHD symptoms and quality of life. AZTEC demonstrates that azacitidine is a safe and promising option for the treatment of cGvHD, and continued evaluation in the second stage of this phase II efficacy study is supported.

Synchronous versus sequential chemo-radiotherapy in patients with early stage breast cancer (SECRAB): A randomised, phase III, trial.

BACKGROUND: The optimal sequence of adjuvant chemotherapy and radiotherapy for breast cancer is unknown. SECRAB assesses whether local control can be improved without increased toxicity. METHODS: SECRAB was a prospective, open-label, multi-centre, phase III trial comparing synchronous to sequential chemo-radiotherapy, conducted in 48 UK centres. Patients with invasive, early stage breast cancer were eligible. Randomisation (performed using random permuted block assignment) was stratified by centre, axillary surgery, chemotherapy, and radiotherapy boost. Permitted chemotherapy regimens included CMF and anthracycline-CMF. Synchronous radiotherapy was administered between cycles two and three for CMF or five and six for anthracycline-CMF. Sequential radiotherapy was delivered on chemotherapy completion. Radiotherapy schedules included 40 Gy/15F over three weeks, and 50 Gy/25F over five weeks. The primary outcome was local recurrence at five and ten years, defined as time to local recurrence, and analysed by intention to treat. ClinicalTrials.gov NCT00003893. FINDINGS: Between 02-July-1998 and 25-March-2004, 2297 patients were recruited (1150 synchronous and 1146 sequential). Baseline characteristics were balanced. With 10.2 years median follow-up, the ten-year local recurrence rates were 4.6% and 7.1% in the synchronous and sequential arms respectively (hazard ratio (HR) 0.62; 95% confidence interval (CI): 0.43-0.90; p = 0.012). In a planned sub-group analysis of anthracycline-CMF, the ten-year local recurrence rates difference were 3.5% versus 6.7% respectively (HR 0.48 95% CI: 0.26-0.88; p = 0.018). There was no significant difference in overall or disease-free survival. 24% of patients on the synchronous arm suffered moderate/severe acute skin reactions compared to 15% on the sequential arm (p < 0.0001). There were no significant differences in late adverse effects apart from telangiectasia (p = 0.03). INTERPRETATION: Synchronous chemo-radiotherapy significantly improved local recurrence rates. This was delivered with an acceptable increase in acute toxicity. The greatest benefit of synchronous chemo-radiation was in patients treated with anthracycline-CMF. FUNDING: Cancer Research UK (CR UK/98/001) and Pharmacia.

HER2 status predicts for upfront AI benefit: A TRANS-AIOG meta-analysis of 12,129 patients from ATAC, BIG 1-98 and TEAM with centrally determined HER2.

BACKGROUND: A meta-analysis of the effects of HER2 status, specifically within the first 2-3 years of adjuvant endocrine therapy, has the potential to inform patient selection for upfront aromatase inhibitor (AI) therapy or switching strategy tamoxifen followed by AI. The pre-existing standardisation of methodology for HER2 (immunohistochemistry/fluorescence in situ hybridization) facilitates analysis of existing data for this key marker. METHODS: Following a prospectively designed statistical analysis plan, patient data from 3 phase III trials Arimidex, Tamoxifen, Alone or in Combination Trial (ATAC), Breast International Group (BIG) 1-98 and Tamoxifen Exemestane Adjuvant Multicentre Trial (TEAM)] comparing an AI to tamoxifen during the first 2-3 years of adjuvant endocrine treatment were collected and a treatment-by-marker analysis of distant recurrence-free interval-censored at 2-3 years treatment - for HER2 status × AI versus tamoxifen treatment was performed to address the clinical question relating to efficacy of 'upfront' versus 'switch' strategies for AIs. RESULTS: A prospectively planned, patient-level data meta-analysis across 3 trials demonstrated a significant treatment (AI versus tamoxifen) by marker (HER2) interaction in a multivariate analysis; (interaction hazard ratio [HR] = 1.61, 95% CI 1.01-2.57; p < 0.05). Heterogeneity between trials did not reach statistical significance. The HER2 negative (HER2-ve) group gained greater benefit from AI versus tamoxifen (HR = 0.70, 95% CI 0.56-0.87) than the HER2-positive (HER2+ve) group (HR = 1.13, 95% CI 0.75-1.71). However, the small number of HER2+ve cases (n = 1092 across the 3 trials) and distant recurrences (n = 111) may explain heterogeneity between trials. CONCLUSIONS: A patient-level data meta-analysis demonstrated a significant interaction between HER2 status and treatment with AI versus tamoxifen in the first 2-3 years of adjuvant endocrine therapy. Patients with HER2-ve cancers experienced improved outcomes (distant relapse) when treated with upfront AI rather than tamoxifen, whilst patients with HER2+ve cancers fared no better or slightly worse in the first 2-3 years. However, the small number of HER2+ve cancers/events may explain a large degree of heterogeneity in the HER2+ve groups across all 3 trials. Other causes, perhaps related to subtle differences between AIs, cannot be excluded and warrant further exploration.

Efficacy, outcomes, and complication rates of different surgical and nonsurgical treatment modalities for recurrent/residual oropharyngeal carcinoma: A systematic review and meta-analysis.

BACKGROUND: Treatment of recurrent oropharyngeal cancer is widely thought to have poor outcomes. Justification for treatment, especially in advanced cases, can be difficult. METHODS: A systematic search of MEDLINE, Embase, and Cochrane databases was conducted. Included studies reported specific recurrent oropharyngeal cancer survival data. RESULTS: Twenty-two retrospective studies were included. Pooled 3-year overall survival (OS) was 26% (95% confidence interval [CI] = 22% to 29%; I squared = 40.7%; p = .057). Pooled 5-year OS was 23% (95% CI = 20% to 27%; I squared = 73.9%; p = .000). Surgical treatment was superior to radiation (5-year OS 26% vs 16%, respectively; p < .001). The 5-year OS improved over time: 18% in the pre-2000 cohort; 35% in the mixed pre-2000 and post-2000 group; and 51% in the post-2000 cohort (p < .001). CONCLUSION: Outcomes have improved considerably over the last 2 decades, resulting in approximately 50% overall 5-year survival. Human papillomavirus (HPV) status, patient selection, and improvements in care may explain this. © 2016 The Authors Head & Neck Published by Wiley Periodicals, Inc. Head Neck, 2016 © 2016 Wiley Periodicals, Inc. Head Neck 38: 1855-1861, 2016.

Mutational analysis of disease relapse in patients allografted for acute myeloid leukemia.

Disease relapse is the major cause of treatment failure after allogeneic stem cell transplantation (allo-SCT) in acute myeloid leukemia (AML). To identify AML-associated genes prognostic of AML relapse post-allo-SCT, we resequenced 35 genes in 113 adults at diagnosis, 49 of whom relapsed. Two hundred sixty-two mutations were detected in 102/113 (90%) patients. An increased risk of relapse was observed in patients with mutations in WT1 (P = .018), DNMT3A (P = .045), FLT3 ITD (P = .071), and TP53 (P = .06), whereas mutations in IDH1 were associated with a reduced risk of disease relapse (P = .018). In 29 patients, we additionally compared mutational profiles in bone marrow at diagnosis and relapse to study changes in clonal structure at relapse. In 13/29 patients, mutational profiles altered at relapse. In 9 patients, mutations present at relapse were not detected at diagnosis. In 15 patients, additional available pre-allo-SCT samples demonstrated that mutations identified posttransplant but not at diagnosis were detectable immediately prior to transplant in 2 of 15 patients. Taken together, these observations, if confirmed in larger studies, have the potential to inform the design of novel strategies to reduce posttransplant relapse highlighting the potential importance of post-allo-SCT interventions with a broad antitumor specificity in contrast to targeted therapies based on mutational profile at diagnosis.

Summarising and validating test accuracy results across multiple studies for use in clinical practice.

Following a meta-analysis of test accuracy studies, the translation of summary results into clinical practice is potentially problematic. The sensitivity, specificity and positive (PPV) and negative (NPV) predictive values of a test may differ substantially from the average meta-analysis findings, because of heterogeneity. Clinicians thus need more guidance: given the meta-analysis, is a test likely to be useful in new populations, and if so, how should test results inform the probability of existing disease (for a diagnostic test) or future adverse outcome (for a prognostic test)? We propose ways to address this. Firstly, following a meta-analysis, we suggest deriving prediction intervals and probability statements about the potential accuracy of a test in a new population. Secondly, we suggest strategies on how clinicians should derive post-test probabilities (PPV and NPV) in a new population based on existing meta-analysis results and propose a cross-validation approach for examining and comparing their calibration performance. Application is made to two clinical examples. In the first example, the joint probability that both sensitivity and specificity will be >80% in a new population is just 0.19, because of a low sensitivity. However, the summary PPV of 0.97 is high and calibrates well in new populations, with a probability of 0.78 that the true PPV will be at least 0.95. In the second example, post-test probabilities calibrate better when tailored to the prevalence in the new population, with cross-validation revealing a probability of 0.97 that the observed NPV will be within 10% of the predicted NPV.

Meta-analysis of test accuracy studies: an exploratory method for investigating the impact of missing thresholds.

BACKGROUND: Primary studies examining the accuracy of a continuous test evaluate its sensitivity and specificity at one or more thresholds. Meta-analysts then usually perform a separate meta-analysis for each threshold. However, the number of studies available for each threshold is often very different, as primary studies are inconsistent in the thresholds reported. Furthermore, of concern is selective reporting bias, because primary studies may be less likely to report a threshold when it gives low sensitivity and/or specificity estimates. This may lead to biased meta-analysis results. We developed an exploratory method to examine the potential impact of missing thresholds on conclusions from a test accuracy meta-analysis. METHODS: Our method identifies studies that contain missing thresholds bounded between a pair of higher and lower thresholds for which results are available. The bounded missing threshold results (two-by-two tables) are then imputed, by assuming a linear relationship between threshold value and each of logit-sensitivity and logit-specificity. The imputed results are then added to the meta-analysis, to ascertain if original conclusions are robust. The method is evaluated through simulation, and application made to 13 studies evaluating protein:creatinine ratio (PCR) for detecting proteinuria in pregnancy with 23 different thresholds, ranging from one to seven per study. RESULTS: The simulation shows the imputation method leads to meta-analysis estimates with smaller mean-square error. In the PCR application, it provides 50 additional results for meta-analysis and their inclusion produces lower test accuracy results than originally identified. For example, at a PCR threshold of 0.16, the summary specificity is 0.80 when using the original data, but 0.66 when also including the imputed data. At a PCR threshold of 0.25, the summary sensitivity is reduced from 0.95 to 0.85 when additionally including the imputed data. CONCLUSIONS: The imputation method is a practical tool for researchers (often non-statisticians) to explore the potential impact of missing threshold results on their meta-analysis conclusions. Software is available to implement the method. In the PCR example, it revealed threshold results are vulnerable to the missing data, and so stimulates the need for advanced statistical models or, preferably, individual patient data from primary studies.

Developing and validating risk prediction models in an individual participant data meta-analysis.

BACKGROUND: Risk prediction models estimate the risk of developing future outcomes for individuals based on one or more underlying characteristics (predictors). We review how researchers develop and validate risk prediction models within an individual participant data (IPD) meta-analysis, in order to assess the feasibility and conduct of the approach. METHODS: A qualitative review of the aims, methodology, and reporting in 15 articles that developed a risk prediction model using IPD from multiple studies. RESULTS: The IPD approach offers many opportunities but methodological challenges exist, including: unavailability of requested IPD, missing patient data and predictors, and between-study heterogeneity in methods of measurement, outcome definitions and predictor effects. Most articles develop their model using IPD from all available studies and perform only an internal validation (on the same set of data). Ten of the 15 articles did not allow for any study differences in baseline risk (intercepts), potentially limiting their model's applicability and performance in some populations. Only two articles used external validation (on different data), including a novel method which develops the model on all but one of the IPD studies, tests performance in the excluded study, and repeats by rotating the omitted study. CONCLUSIONS: An IPD meta-analysis offers unique opportunities for risk prediction research. Researchers can make more of this by allowing separate model intercept terms for each study (population) to improve generalisability, and by using 'internal-external cross-validation' to simultaneously develop and validate their model. Methodological challenges can be reduced by prospectively planned collaborations that share IPD for risk prediction.

A framework for developing, implementing, and evaluating clinical prediction models in an individual participant data meta-analysis.

The use of individual participant data (IPD) from multiple studies is an increasingly popular approach when developing a multivariable risk prediction model. Corresponding datasets, however, typically differ in important aspects, such as baseline risk. This has driven the adoption of meta-analytical approaches for appropriately dealing with heterogeneity between study populations. Although these approaches provide an averaged prediction model across all studies, little guidance exists about how to apply or validate this model to new individuals or study populations outside the derivation data. We consider several approaches to develop a multivariable logistic regression model from an IPD meta-analysis (IPD-MA) with potential between-study heterogeneity. We also propose strategies for choosing a valid model intercept for when the model is to be validated or applied to new individuals or study populations. These strategies can be implemented by the IPD-MA developers or future model validators. Finally, we show how model generalizability can be evaluated when external validation data are lacking using internal-external cross-validation and extend our framework to count and time-to-event data. In an empirical evaluation, our results show how stratified estimation allows study-specific model intercepts, which can then inform the intercept to be used when applying the model in practice, even to a population not represented by included studies. In summary, our framework allows the development (through stratified estimation), implementation in new individuals (through focused intercept choice), and evaluation (through internal-external validation) of a single, integrated prediction model from an IPD-MA in order to achieve improved model performance and generalizability.

Assessment of publication bias, selection bias, and unavailable data in meta-analyses using individual participant data: a database survey.

OBJECTIVE: To examine the potential for publication bias, data availability bias, and reviewer selection bias in recently published meta-analyses that use individual participant data and to investigate whether authors of such meta-analyses seemed aware of these issues. DESIGN: In a database of 383 meta-analyses of individual participant data that were published between 1991 and March 2009, we surveyed the 31 most recent meta-analyses of randomised trials that examined whether an intervention was effective. Identification of relevant articles and data extraction was undertaken by one author and checked by another. RESULTS: Only nine (29%) of the 31 meta-analyses included individual participant data from "grey literature" (such as unpublished studies) in their primary meta-analysis, and the potential for publication bias was discussed or investigated in just 10 (32%). Sixteen (52%) of the 31 meta-analyses did not obtain all the individual participant data requested, yet five of these (31%) did not mention this as a potential limitation, and only six (38%) examined how trials without individual participant data might affect the conclusions. In nine (29%) of the meta-analyses reviewer selection bias was a potential issue, as the identification of relevant trials was either not stated or based on a more selective, non-systematic approach. Investigation of four meta-analyses containing data from ≥10 trials revealed one with an asymmetric funnel plot consistent with publication bias, and the inclusion of studies without individual participant data revealed additional heterogeneity between trials. CONCLUSIONS: Publication, availability, and selection biases are a potential concern for meta-analyses of individual participant data, but many reviewers neglect to examine or discuss them. These issues warn against uncritically viewing any meta-analysis that uses individual participant data as the most reliable. Reviewers should seek individual participant data from all studies identified by a systematic review; include, where possible, aggregate data from any studies lacking individual participant data to consider their potential impact; and investigate funnel plot asymmetry in line with recent guidelines.

A systematic review of breast cancer incidence risk prediction models with meta-analysis of their performance.

A risk prediction model is a statistical tool for estimating the probability that a currently healthy individual with specific risk factors will develop a condition in the future such as breast cancer. Reliably accurate prediction models can inform future disease burdens, health policies and individual decisions. Breast cancer prediction models containing modifiable risk factors, such as alcohol consumption, BMI or weight, condom use, exogenous hormone use and physical activity, are of particular interest to women who might be considering how to reduce their risk of breast cancer and clinicians developing health policies to reduce population incidence rates. We performed a systematic review to identify and evaluate the performance of prediction models for breast cancer that contain modifiable factors. A protocol was developed and a sensitive search in databases including MEDLINE and EMBASE was conducted in June 2010. Extensive use was made of reference lists. Included were any articles proposing or validating a breast cancer prediction model in a general female population, with no language restrictions. Duplicate data extraction and quality assessment were conducted. Results were summarised qualitatively, and where possible meta-analysis of model performance statistics was undertaken. The systematic review found 17 breast cancer models, each containing a different but often overlapping set of modifiable and other risk factors, combined with an estimated baseline risk that was also often different. Quality of reporting was generally poor, with characteristics of included participants and fitted model results often missing. Only four models received independent validation in external data, most notably the 'Gail 2' model with 12 validations. None of the models demonstrated consistently outstanding ability to accurately discriminate between those who did and those who did not develop breast cancer. For example, random-effects meta-analyses of the performance of the 'Gail 2' model showed the average C statistic was 0.63 (95% CI 0.59-0.67), and the expected/observed ratio of events varied considerably across studies (95% prediction interval for E/O ratio when the model was applied in practice was 0.75-1.19). There is a need for models with better predictive performance but, given the large amount of work already conducted, further improvement of existing models based on conventional risk factors is perhaps unlikely. Research to identify new risk factors with large additionally predictive ability is therefore needed, alongside clearer reporting and continual validation of new models as they develop.