Long-term behavioral symptom clusters among survivors of early-stage breast cancer. development and validation of a predictive model.

BACKGROUND: Fatigue, cognitive impairment, anxiety, depression, and sleep disturbance are cancer-related behavioral symptoms (CRBS) that may persist years after early-stage breast cancer (BC), affecting quality of life. We aimed at generating a predictive model of long-term CRBS clusters among BC survivors four years post-diagnosis. METHODS: Patients with early-stage BC were included from the CANcer TOxicity (NCT01993498). Our outcome was the proportion of patients reporting CRBS clusters four years post-diagnosis (≥3 severe CRBS). Predictors, including clinical, behavioral, and treatment-related characteristics, Behavioral Symptoms Score (BSS; 1 point per severe CRBS at diagnosis) and a pro-inflammatory cytokine (IL-1b, IL-6, TNFα)-genetic risk score, were tested using multivariable logistic regression, implementing bootstrapped Augmented Backwards Elimination. A two-sided p-value < 0.05 defined statistical significance. RESULTS: In the development cohort (N = 3555), 642 patients (19.0%) reported a cluster of CRBS at diagnosis and 755 (21.2%) did so four years post-diagnosis. Younger age (adjusted Odds Ratio [aOR] for 1-year decrement: 1.012; 95% Confidence Interval [CI] 1.003-1.020); previous psychiatric disorders (aOR vs no: 1.27; 95% CI 1.01-1.60); and BSS (aOR ranged from 2.17 [1.66-2.85] for BSS = 1 vs 0 to 12.3 [7.33-20.87] for BSS = 5 vs 0) were predictors of reporting a cluster of CRBS (AUC 0.73 [95%CI 0.71-0.75]). Genetic risk score was not predictive of CRBS. Results were confirmed in the validation cohort (N = 1533). CONCLUSION: Younger patients with previous psychiatric disorders and higher baseline symptom burden have greater risk of long-term clusters of CRBS. Our model might be implemented in clinical pathways to improve management and test the effectiveness of risk mitigation interventions among breast cancer survivors.

Metabolomic Prediction of Breast Cancer Treatment-Induced Neurologic and Metabolic Toxicities.

PURPOSE: Long-term treatment-related toxicities, such as neurologic and metabolic toxicities, are major issues in breast cancer. We investigated the interest of metabolomic profiling to predict toxicities. EXPERIMENTAL DESIGN: Untargeted high-resolution metabolomic profiles of 992 patients with estrogen receptor (ER)+/HER2- breast cancer from the prospective CANTO cohort were acquired (n = 1935 metabolites). A residual-based modeling strategy with discovery and validation cohorts was used to benchmark machine learning algorithms, taking into account confounding variables. RESULTS: Adaptive Least Absolute Shrinkage and Selection (adaptive LASSO) has a good predictive performance, has limited optimism bias, and allows the selection of metabolites of interest for future translational research. The addition of low-frequency metabolites and nonannotated metabolites increases the predictive power. Metabolomics adds extra performance to clinical variables to predict various neurologic and metabolic toxicity profiles. CONCLUSIONS: Untargeted high-resolution metabolomics allows better toxicity prediction by considering environmental exposure, metabolites linked to microbiota, and low-frequency metabolites.

An Early Neoplasia Index (ENI10), Based on Molecular Identity of CD10 Cells and Associated Stemness Biomarkers, is a Predictor of Patient Outcome in Many Cancers.

An accurate estimate of patient survival at diagnosis is critical to plan efficient therapeutic options. A simple and multiapplication tool is needed to move forward the precision medicine era. Taking advantage of the broad and high CD10 expression in stem and cancers cells, we evaluated the molecular identity of aggressive cancer cells. We used epithelial primary cells and developed a breast cancer stem cell­based progressive model. The superiority of the early-transformed isolated molecular index was evaluated by large-scale analysis in solid cancers. BMP2-driven cell transformation increases CD10 expression which preserves stemness properties. Our model identified a unique set of 159 genes enriched in G2­M cell-cycle phases and spindle assembly complex. Using samples predisposed to transformation, we confirmed the value of an early neoplasia index associated to CD10 (ENI10) to discriminate premalignant status of a human tissue. Using a stratified Cox model, a large-scale analysis (>10,000 samples, The Cancer Genome Atlas Pan-Cancer) validated a strong risk gradient (HRs reaching HR = 5.15; 95% confidence interval: 4.00­6.64) for high ENI10 levels. Through different databases, Cox regression model analyses highlighted an association between ENI10 and poor progression-free intervals for more than 50% of cancer subtypes tested, and the potential of ENI10 to predict drug efficacy. The ENI10 index constitutes a robust tool to detect pretransformed tissues and identify high-risk patients at diagnosis. Owing to its biological link with refractory cancer stem cells, the ENI10 index constitutes a unique way of identifying effective treatments to improve clinical care. SIGNIFICANCE: We identified a molecular signature called ENI10 which, owing to its biological link with stem cell properties, predicts patient outcome and drugs efficiency in breast and several other cancers. ENI10 should allow early and optimized clinical management of a broad number of cancers, regardless of the stage of tumor progression.

Nuclear PRMT5 is a biomarker of sensitivity to tamoxifen in ERα+ breast cancer.

Endocrine therapies targeting estrogen signaling, such as tamoxifen, have significantly improved management of estrogen receptor alpha (ERα)-positive breast cancers. However, their efficacy is limited by intrinsic and acquired resistance to treatment, and there is currently no predictive marker of response to these anti-estrogens to guide treatment decision. Here, using two independent cohorts of breast cancer patients, we identified nuclear PRMT5 expression as an independent predictive marker of sensitivity to tamoxifen. Mechanistically, we discovered that tamoxifen stimulates ERα methylation by PRMT5, a key event for its binding to corepressors such as SMRT and HDAC1, participating in the inhibition of the transcriptional activity of ERα. Although PRMT5 is mainly localized in the cytoplasm of tumor cells, our analyses show that tamoxifen triggers its nuclear translocation in tamoxifen-sensitive tumors but not in resistant ones. Hence, we unveil a biomarker of sensitivity to tamoxifen in ERα-positive breast tumors that could be used to enhance the response of breast cancer patients to endocrine therapy, by fostering its nuclear expression.

Target trial emulation to assess real-world efficacy in the Epidemiological Strategy and Medical Economics metastatic breast cancer cohort.

BACKGROUND: Real-world data studies usually consider biases related to measured confounders. We emulate a target trial implementing study design principles of randomized trials to observational studies; controlling biases related to selection, especially immortal time; and measured confounders. METHODS: This comprehensive analysis emulating a randomized clinical trial compared overall survival in patients with HER2-negative metastatic breast cancer (MBC), receiving as first-line treatment, either paclitaxel alone or combined to bevacizumab. We used data from 5538 patients extracted from the Epidemiological Strategy and Medical Economics-MBC cohort to emulate a target trial using advanced statistical adjustment techniques including stabilized inverse-probability weighting and G-computation, dealing with missing data with multiple imputation, and performing a quantitative bias analysis for residual bias due to unmeasured confounders. RESULTS: Emulation led to 3211 eligible patients, and overall survival estimates achieved with advanced statistical methods favored the combination therapy. Real-world effect sizes were close to that assessed in the existing E2100 randomized clinical trial (hazard ratio = 0.88, P = .16), but the increased sample size allowed to achieve a higher level of precision in real-world estimates (ie, reduced confidence intervals). Quantitative bias analysis confirmed the robustness of the results with respect to potential unmeasured confounding. CONCLUSION: Target trial emulation with advanced statistical adjustment techniques is a promising approach to investigate long-term impact of innovative therapies in the French Epidemiological Strategy and Medical Economics-MBC cohort while minimizing biases and provides opportunities for comparative efficacy through the synthetic control arms provided. DATABASE REGISTRATION: clinicaltrials.gov Identifier NCT03275311.

First estimates of diffuse gastric cancer risks for carriers of CTNNA1 germline pathogenic variants.

BACKGROUND: Pathogenic variants (PV) of CTNNA1 are found in families fulfilling criteria for hereditary diffuse gastric cancer (HDGC) but no risk estimates were available until now. The aim of this study is to evaluate diffuse gastric cancer (DGC) risks for carriers of germline CTNNA1 PV. METHODS: Data from published CTNNA1 families were updated and new families were identified through international collaborations. The cumulative risk of DGC by age for PV carriers was estimated with the genotype restricted likelihood (GRL) method, taking into account non-genotyped individuals and conditioning on all observed phenotypes and genotypes of the index case to obtain unbiased estimates. A non-parametric (NP) and the Weibull functions were used to model the shape of penetrance function with the GRL. Kaplan-Meier incidence curve and standardised incidence ratios were also computed. A 'leave-one-out' strategy was used to evaluate estimate uncertainty. RESULTS: Thirteen families with 46 carriers of PV were included. The cumulative risks of DGC at 80 years for carriers of CTNNA1 PV are 49% and 57%, respectively with the Weibull GRL and NP GRL methods. Risk ratios to population incidence reach particularly high values at early ages and decrease with age. At 40 years, they are equal to 65 and 833, respectively with the Weibull GRL and NP GRL. CONCLUSION: This is the largest series of CTNNA1 families that provides the first risk estimates of GC. These data will help to improve management and surveillance for these patients and support inclusion of CTNNA1 in germline testing panels.

Anti-müllerian hormone levels and antral follicle count in women with a BRCA1 or BRCA2 germline pathogenic variant: A retrospective cohort study.

BACKGROUND: Some studies suggested a decreased ovarian reserve among BRCA1/2 pathogenic variant carriers, with conflicting results. METHODS: We conducted a retrospective single-center observational study of ovarian reserve and spontaneous fertility comparing BRCA1/2 pathogenic variant carriers to controls (women who attended consultations to discuss fertility preservation before gonadotoxic treatment). Measures of associations between plasma AMH concentration, AFC and BRCA1/2 status were modelled by nonlinear generalized additive regression models and logistic regressions adjusted for age at plasma storage, oral contraceptive use, body mass index, cigarette smoking, and the AMH assay technique. RESULTS: The whole population comprised 119 BRCA1/2 pathogenic variant carriers and 92 controls. A total of 110 women (42 carriers, among whom 30 were cancer-free, and 68 controls) underwent an ovarian reserve evaluation. Spontaneous fertility analysis included all women who previously attempted to become pregnant (134 women). We observed a tendency towards a premature decrease in ovarian reserve in BRCA1/2 pathogenic variant carriers, but no difference in mean AMH or AFC levels was found between BRCA1/2 pathogenic variant carriers and controls. An analysis of the extreme levels of AMH (≤5 pmol/l) and AFC (≤7 follicles) by logistic regression suggested a higher risk of low ovarian reserve among BRCA1/2 pathogenic variant carriers (adjusted odds ratio (OR) = 3.57, 95% CI = 1.00-12.8, p = 0.05; and adjusted OR = 4.99, 95% CI = 1.10-22.62, p = 0.04, respectively). DISCUSSION: Attention should be paid to BRCA1/2 pathogenic variant carriers' ovarian reserve, considering this potential risk of premature alteration.

Integrated analysis highlights APC11 protein expression as a likely new independent predictive marker for colorectal cancer.

After a diagnosis of colorectal cancer (CRC), approximately 50% of patients will present distant metastasis. Although significant progress has been made in treatments, most of them will die from the disease. We investigated the predictive and prognostic potential of APC11, the catalytic subunit of APC/C, which has never been examined in the context of CRC. The expression of APC11 was assessed in CRC cell lines, in tissue microarrays (TMAs) and in public datasets. Overexpression of APC11 mRNA was associated with chromosomal instability, lymphovascular invasion and residual tumor. Regression models accounting for the effects of well-known protein markers highlighted association of APC11 protein expression with residual tumor (odds ratio: OR = 6.51; 95% confidence intervals: CI = 1.54-27.59; P = 0.012) and metastasis at diagnosis (OR = 3.87; 95% CI = 1.20-2.45; P = 0.024). Overexpression of APC11 protein was also associated with worse distant relapse-free survival (hazard ratio: HR = 2.60; 95% CI = 1.26-5.37; P = 0.01) and worse overall survival (HR = 2.69; 95% CI = 1.31-5.51; P = 0.007). APC11 overexpression in primary CRC thus represents a potentially novel theranostic marker of metastatic CRC.

Short-term risk of colorectal cancer in individuals with lynch syndrome: a meta-analysis.

PURPOSE: For carriers of germline mutations in DNA mismatch repair genes, the most relevant statistic for cancer prevention is colorectal cancer (Lynch syndrome) risk, particularly in the short term. METHODS: We conducted a meta-analysis of all independent published Lynch syndrome studies reporting age- and sex-dependent colorectal cancer risks. We estimated 5-year colorectal cancer risk over different age groups, separately for male and female mutation carriers, and number needed to screen to prevent one death. RESULTS: We pooled estimates from analyses of 1,114 Lynch syndrome families (508 with MLH1 mutations and 606 with MSH2 mutations). On average, one in 71 male and one in 102 female MLH1 or MSH2 mutation carriers in their 20s will be diagnosed with colorectal cancer in the next 5 years. These colorectal cancer risks increase with age, peaking in the 50s (one in seven males and one in 12 females), and then decrease with age (one in 13 males and one in 19 females in their 70s). Annual colonoscopy in 16 males or 25 females in their 50s would prevent one death from colorectal cancer over 5 years while resulting in almost no serious complications. In comparison, annual colonoscopy in 155 males or 217 females in their 20s would prevent one death while resulting in approximately one serious complication. CONCLUSION: For MLH1 or MSH2 mutation carriers, current guidelines recommend colonoscopy every 1 to 2 years starting in their 20s. Our findings support this regimen from age 30 years; however, it might not be justifiable for carriers who are in their 20s.

Management of patients diagnosed with atypical ductal hyperplasia by vacuum-assisted core biopsy: a prospective assessment of the guidelines used at our institution.

BACKGROUND: Because of underestimation, surgical excision is recommended for atypical ductal hyperplasia diagnosed on directional vacuum-assisted biopsies. The following guidelines have been established according to our retrospective study published in 2008: excision for lesions ≥ 21 mm, follow-up for lesions <6 mm with complete removal of microcalcifications, and follow-up or excision for 6 to 21-mm lesions with respectively less or >2 atypical ductal hyperplasia foci. METHODS AND RESULTS: These guidelines were assessed in a prospective series of 124 patients with a median follow-up of 30 months. Conformity rate was 92%. Upgrading was 28% (15 of 53 patients) for conformed surgery and absent for surgery performed beyond the scope of guidelines. For the patients with benign result at surgery (n = 38) or just followed (n = 61), 3 cancers occurred in either breast at 1 to 3 years. CONCLUSIONS: These convenient guidelines can safely spare surgery for a subset of patients. However, annual mammographic follow-up is recommended since the risk of subsequent cancer remains high for both breasts.

Evidence of time-dependent prognostic factors predicting early death but not long-term outcome in primary CNS lymphoma: a study of 91 patients.

Long-term primary CNS lymphoma (PCNSL) survivors could have multiple adverse prognostic factors at diagnosis such as age, performance status (PS), site of the tumour (deep vs superficial), lactate dehydrogenase (LDH) level and CSF protein level. Whether these five prognostic factors integrated in the International Extranodal Lymphoma Study Group (IELSG) score have a time-dependent effect is questionable. Among 132 PCNSL patients treated at our institution between 1984 and 2006, 91 available patients for IELSG score were evaluated by time-segmented analysis. Of the 91 patients, 21% had 0-1, 59% had 2-3 and 20% had 4-5 adverse IELSG prognostic scores. With a median follow-up of 102 months, the median overall survival (OS) of the 91 patients with the five prognostic factors of IELSG score was 33 months (95% CI, 17 to 55) compared with 14 months (95% CI, 3 to 23) for the remaining 41 patients whose CSF protein level was lacking in the IELSG score. These 41 patients who did not have lumbar puncture presented a poorer PS at diagnosis and a lower treatment response rate. While confirming the prognostic value of the IELSG score, we observed a time-dependent effect of age, PS and tumour site; all three lost their prognostic value after 6 months from diagnosis, while LDH remained a constant predictor of OS. No prognostic impact of CSF protein level was reported. Patients with older age, poor PS and deep brain involvement are at risk of death during the first months after diagnosis but could have a favourable long-term outcome after the treatment period. New prognostic factors predicting long-term outcome remain to be determined.

Mutational characterization of individual breast tumors: TP53 and PI3K pathway genes are frequently and distinctively mutated in different subtypes.

Understanding how cancer genes are mutated in individual tumors is an important issue with potential clinical and therapeutic impact. This is especially relevant with recently developed targeted therapies since mutated genes can be targets and/or predictors. However, to date, gene mutation profiling in individual tumors is still underexplored. Breast cancer is composed of various subtypes. We presumed that this heterogeneity reflected the involvement of different molecular mechanisms including gene mutations that affect defined signaling pathways. Unlike the majority of published mutational studies, this study was aimed to draw a mutation profile in individual tumors by screening a panel of cancer genes in the same tumor. Thus, five genes frequently mutated in breast cancers: TP53, PIK3CA, PTEN, CDH1, and AKT1 were screened in each of 120 human primary breast tumors. Mutations in at least one of these genes were found in 62.5% of the tumors, of which the majority carried a single-gene mutation. Interestingly, a substantial proportion of tumors carried mutations either in TP53 or in genes of the PI3K pathway (PIK3CA or PTEN or AKT1). These two distinct mutation patterns were significantly associated to hormone receptor expression but independent of HER2 status.

Variants in the netrin-1 receptor UNC5C prevent apoptosis and increase risk of familial colorectal cancer.

BACKGROUND & AIMS: Expression of the netrin-1 dependence receptor UNC5C is reduced in many colorectal tumors; mice with the UNC5C mutations have increased progression of intestinal tumors. We investigated whether specific variants in UNC5C increase risk of colorectal cancer (CRC). METHODS: We analyzed the sequence of UNC5C in blood samples from 1801 patients with CRC and 4152 controls from 3 cohorts (France, United States, and Finland). Almost all cases from France and the United States had familial CRC; of the Finnish cases, 92 of 984 were familial. We analyzed whether CRC segregates with the UNC5C variant A628K in 3 families with histories of CRC. We also performed haplotype analysis to determine the origin of this variant. RESULTS: Of 817 patients with familial CRC, 14 had 1 of 4 different, unreported missense variants in UNC5C. The variants p.Asp353Asn (encodes D353N), p.Arg603Cys (encodes R603C), and p.Gln630Glu (encodes Q630E) did not occur significantly more often in cases than controls. The variant p.Ala628Lys (A628K) was detected in 3 families in the French cohort (odds ratio, 8.8; Wald's 95% confidence interval, 1.47-52.93; P = .03) and in 2 families in the US cohort (odds ratio, 1.9; P = .6) but was not detected in the Finnish cohort; UNC5C A628K segregated with CRC in families. Three families with A628K had a 109-kilobase identical haplotype that spanned most of UNC5C, indicating recent origin of this variant in white subjects (14 generations; 95% confidence interval, 6-36 generations). Transfection of HEK293T cells with UNC5C-A628K significantly reduced apoptosis compared with wild-type UNC5C, measured in an assay of active caspase-3. CONCLUSIONS: Inherited mutations in UNC5C prevent apoptosis and increase risk of CRC.

Central venous catheter database: an important issue in quality assurance.

BACKGROUND/PURPOSE: The purpose of this study was to analyze the factors that affect the longevity of central venous catheters. METHODS: Comprehensive clinical data recorded during insertion and removal of totally implantable devices (TID) and tunneled lines (TL) from October 1988 to January 2009 were analyzed. Univariate and multivariate Cox proportional hazards regression models were used to identify clinical factors that predict catheter longevity. RESULTS: Information was available for 1167 central venous catheter insertions in 858 patients, 648 TID and 509 TL. Univariate analysis detected longer device longevity in the following: TID longer than TL (P < .0001), catheter tip in the superior vena cava (SVC)/right atrial junction (P < .0001), and right side greater than left (P = .002). Shorter device longevity was observed in lines used for total parenteral nutrition (P < .0001) and young age (P < .0001). Multivariate model detected the following: hazard of removal for TID is 0.304 that of TL (P < .0001) and SVC is 0.525 that of other locations (P = .0005). Hazard decreases by 5.4% for every 1-year increase in patient age (P < .0004). CONCLUSION: Multiple confounding factors were encountered. However, the single most important factor in catheter longevity that is influenced by the surgeon is tip location in the SVC/right atrial junction.

Evaluating common de-identification heuristics for personal health information.

BACKGROUND: With the growing adoption of electronic medical records, there are increasing demands for the use of this electronic clinical data in observational research. A frequent ethics board requirement for such secondary use of personal health information in observational research is that the data be de-identified. De-identification heuristics are provided in the Health Insurance Portability and Accountability Act Privacy Rule, funding agency and professional association privacy guidelines, and common practice. OBJECTIVE: The aim of the study was to evaluate whether the re-identification risks due to record linkage are sufficiently low when following common de-identification heuristics and whether the risk is stable across sample sizes and data sets. METHODS: Two methods were followed to construct identification data sets. Re-identification attacks were simulated on these. For each data set we varied the sample size down to 30 individuals, and for each sample size evaluated the risk of re-identification for all combinations of quasi-identifiers. The combinations of quasi-identifiers that were low risk more than 50% of the time were considered stable. RESULTS: The identification data sets we were able to construct were the list of all physicians and the list of all lawyers registered in Ontario, using 1% sampling fractions. The quasi-identifiers of region, gender, and year of birth were found to be low risk more than 50% of the time across both data sets. The combination of gender and region was also found to be low risk more than 50% of the time. We were not able to create an identification data set for the whole population. CONCLUSIONS: Existing Canadian federal and provincial privacy laws help explain why it is difficult to create an identification data set for the whole population. That such examples of high re-identification risk exist for mainstream professions makes a strong case for not disclosing the high-risk variables and their combinations identified here. For professional subpopulations with published membership lists, many variables often needed by researchers would have to be excluded or generalized to ensure consistently low re-identification risk. Data custodians and researchers need to consider other statistical disclosure techniques for protecting privacy.