P5706Finding predictors and causes of cardiac surgery ICU readmission using machine learning and causal inference

Background

“Bounce-back” to the intensive care unit (ICU) occurs when patients return to the ICU for critical changes in clinical status within the same hospital admission. Bounce-backs post-cardiac surgery increase resource utilisation, total cost of care, are associated with higher mortality and morbidity. However, prediction of bounce-back has proved to be challenging. Previous work addressed the feasibility of predicting bounce-back, but these models required significant physician input to design and calibrate the predictive variables.

Purpose

We aimed to develop an automated machine learning model that would identify patients at risk of bounce-back by selecting the most relevant variables from those available before onset of bounce-back. Additionally, we highlight the differences between predictive and causal inference, to demonstrate that purely associative methods of prediction can mislead clinical decision-making.

Methods

Clinical records of adult cardiac surgery patients between 2011 to 2016 were collected from our institutional Society for Thoracic Surgeons (STS) database and our institutional electronic health record (EHR) system. For bounce-back prediction, an L1 regularised logistic regression model was applied, which also automatically determined important variables with highest prediction effect from the initial 151 variables. For causal inference, the g-computation algorithm was used to compare the differences between causal and predictive regression effects. We quantified the performance of our system on clinically relevant metrics such as specificity, sensitivity, and area under the ROC curve (AUC).

Results

Of the 6189 patients, 357 (5.7%) bounced back to the ICU. The prediction model achieved an AUC score of 0.75 (0.03) and 22% specificity at 95% sensitivity, Further analysis showed 79% of the false positive patients had faced other severe postoperative complications but none of the false negative patients had downstream complications. Subsequent causal analysis revealed that the actual causal effects of treatments differed from the predictive model estimates, e.g. administration of intra-operative tranexamic acid increased the probability of bounce-back by 13% but its causal effect on bounce-back after removing confounders was negligible (an increase of only 0.5%).

Conclusions

Our predictive machine-learning model can successfully predict patients at risk of ICU bounce-backs, using linked STS registry data with the comprehensive electronic health record. The prediction model automatically detects important subset of variables. In addition, we note that causal and predictive model estimates of the same parameters differed, indicating that reliance on predictive models for interventional clinical decision-making may not be appropriate.

Acknowledgement/Funding

National Institutes of Health, Office of Naval Research, Defense Advanced Research Projects Agency

Abstract OT2-07-09: A phase Ib study of neratinib, pertuzumab, and trastuzumab with paclitaxel in patients with metastatic and locally advanced breast cancer

Background: Neratinib, a potent irreversible pan-ErbB tyrosine kinase inhibitor that blocks signal transduction through HER1, HER2, and HER4, has demonstrated activity against metastatic HER2-positive breast cancer (HER2+ BC) in patients pretreated with trastuzumab. The FDA recently approved neratinib as an adjuvant treatment for HER2+ BC patients who have received trastuzumab for at least 1 year. Preclinical data demonstrate that trastuzumab-resistant BC cells remain sensitive to neratinib. Also, neratinib enhances responses to trastuzumab possibly by increasing trastuzumab's internalization, immune-mediated action, and other mechanisms. Taken together, these findings provide the rationale for adding neratinib to the standard of care combination of trastuzumab and pertuzumab with paclitaxel to enhance anti-HER2 efficacy in advanced HER2+ BC. Here, we report on the phase Ib portion of an ongoing phase Ib/II trial of this drug combination.

Trial Design: Patients with metastatic or locally advanced HER2+ BC will be enrolled in the phase Ib portion of the trial. Neratinib is given orally in 3-week cycles. The initial neratinib dose of 80 mg daily is increased to 120, 160, and 200 mg daily after safety assessments of each dose level. Other agents are administered as per the standard of care. Patients continue therapy with per-protocol dose escalation and de-escalation according to toxicity until the maximum tolerated dose (MTD) of neratinib is reached. The target maximum dose-limiting toxicity rate is 20%. All patients receive 4 cycles of the combination therapy. If patients do not have disease progression or excessive toxicity, they may receive 2-4 additional cycles at the treating physician's discretion. During therapy, patients undergo blood tests every week and have clinical visits and restaging scans every 3 weeks. Because gastrointestinal toxicity, mainly diarrhea, is anticipated, patients receive prophylactic antidiarrheal medication (e.g., loperamide, budesonide) beginning with the first dose of neratinib.

Eligibility Criteria: Eligible patients must have histologically confirmed metastatic or locally advanced HER2+ BC (BC may be inflammatory or non-inflammatory and have any hormone receptor status); an ECOG performance status score of 0 or 1; and adequate hematologic and organ function, including adequate cardiac function (as indicated by a left ventricle ejection fraction of ≥50%).

Specific Aims:

1- To determine the MTD of neratinib in combination with paclitaxel, pertuzumab, and trastuzumab.

2- Pharmacodynamic markers will be measured on biologic specimens. Neratinib-induced changes in pEGFR and/or HER2 expression will be analyzed and compared between dose levels.

Statistical Methods: The Bayesian modified toxicity probability interval is used to determine dose adjustment.

Accrual: The target enrollment for the phase Ib cohort is 20 patients. The trial has enrolled 3 patients since its activation in January 2018. This trial is supported by Puma Biotechnology, Aggressive Breast Cancer Research Program Grant.

Citation Format: Al-Awadhi A, Kono M, Marx A, Moseley T, Willey J, Sun H, Fu M, Whitman GJ, Valero V, Ueno NT, Lim B. A phase Ib study of neratinib, pertuzumab, and trastuzumab with paclitaxel in patients with metastatic and locally advanced breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr OT2-07-09.

Abstract OT1-01-05: A phase II study using talimogene laherparepvec as a single agent for inflammatory breast cancer or non-inflammatory breast cancer patients with inoperable local recurrence

Objective: The primary purpose of the study is to determine the local and systemic antitumor efficacy of talimogene laherparepvec in locally recurrent breast cancer patients with or without distant metastases, as evidenced by improved overall response rates. This will be the first study to use biopsy of distant disease to demonstrate whether systemic immune modulation has antitumor efficacy in breast cancer patients.

BACKGROUND: Patients with locally recurrent breast disease frequently undergo multimodal treatment at the first occurrence of breast cancer, and because local treatment modalities such as surgical intervention and radiation are difficult to add, they subsequently receive systemic therapy. Talimogene laherparepvec (T-VEC) was developed to eliminate solid tumors and has since been considered as a potential treatment option for body surface tumors. In addition to T-VECinjected area, this agent is capable of modifying the immune response with the potential of inhibiting distant metastases. Hence, locally recurrent breast disease could benefit from T-VECregardless of concomitant distant metastases, and may offer a new local treatment option.

Study Design and Treatment Plan: This is a single agent phase II study. Patients with breast cancer who have recurrence of chest wall disease with or without distant metastasis, have at least 1 injectable lesion ≥5 mm in longest diameter or multiple injectable lesions that in aggregate have a longest diameter of ≥ 5 mm, and meet inclusion and exclusion criteria will be eligible to participate in the study. Patient will receive T-VEC via intra-tumoral injection every 2 weeks after the first initial injection (3 weeks).

STATISTICAL METHODS:

Up to 35 patients will be enrolled in the study. The trial will be conducted using a two-stage design and the overall response rate will be estimated accordingly. It is assumed that the talimogene laherparepvec single agent will have a response rate of 20%. A response rate of 5% or lower will be considered treatment failure and the regimen will be rejected under this circumstance.

Status of the study:

Activation Date: Aug 2016. 6 patients have been treated. Enrollment continues.

Sponsor: Amgen

State of Texas appropriation for rare and aggressive breast cancer research.

Citation Format: Willey JS, Marx AN, Lim B, Ibrahim NK, Valero V, Mittendorf EA, Reuben JM, Le-Petross HT, Whitman GJ, Krishnamurthy S, Woodward WA, Lucci A, Liu DD, Shen Y, Ueno NT. A phase II study using talimogene laherparepvec as a single agent for inflammatory breast cancer or non-inflammatory breast cancer patients with inoperable local recurrence [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr OT1-01-05.

A feasibility study of neoadjuvant talazoparib for operable breast cancer patients with a germline BRCA mutation demonstrates marked activity

This study was undertaken to determine the feasibility of enrolling breast cancer patients on a single-agent-targeted therapy trial before neoadjuvant chemotherapy. Specifically, we evaluated talazoparib in patients harboring a deleterious BRCA mutation (BRCA+). Patients with a germline BRCA mutation and ≥1 cm, HER2-negative primary tumors were eligible. Study participants underwent a pretreatment biopsy, 2 months of talazoparib, off-study core biopsy, anthracycline, and taxane-based chemotherapy ± carboplatin, followed by surgery. Volumetric changes in tumor size were determined by ultrasound at 1 and 2 months of therapy. Success was defined as 20 patients accrued within 2 years and <33% experienced a grade 4 toxicity. The study was stopped early after 13 patients (BRCA1 + n = 10; BRCA2 + n = 3) were accrued within 8 months with no grade 4 toxicities and only one patient requiring dose reduction due to grade 3 neutropenia. The median age was 40 years (range 25–55) and clinical stage included I (n = 2), II (n = 9), and III (n = 2). Most tumors (n = 9) were hormone receptor-negative, and one of these was metaplastic. Decreases in tumor volume occurred in all patients following 2 months of talazoparib; the median was 88% (range 30–98%). Common toxicities were neutropenia, anemia, thrombocytopenia, nausea, dizziness, and fatigue. Single-agent-targeted therapy trials are feasible in BRCA+ patients. Given the rapid rate of accrual, profound response and favorable toxicity profile, the feasibility study was modified into a phase II study to determine pathologic complete response rates after 4–6 months of single-agent talazoparib.

An investigational PARP inhibitor seems safe and possibly effective when given ahead of surgery to women with BRCA-mutated breast cancer. Jennifer Litton and colleagues from the University of Texas MD Anderson Cancer Center in Houston, USA, conducted a small feasibility study to see if patients with stage I-III breast cancer and inherited mutations in BRCA1 or BRCA2 would put off their standard course of chemotherapy ahead of surgery to first receive two months of talazoparib, an experimental inhibitor of poly ADP ribose polymerase (PARP), an enzyme involved in DNA damage repair. The trial was a resounding success. In fact, owing to rapid patient enrollment, decreases in tumor volume among all 13 participants and few signs of serious side effects, the researchers amended the study protocol to give talazoparib for longer and test for therapeutic efficacy.

Abstract P1-14-04: A randomized phase II neoadjuvant (NACT) study of sequential eribulin followed by FAC/FEC-regimen compared to sequential paclitaxel followed by FAC/FEC-regimen in patients (pts) with operable breast cancer not overexpressing HER-2

Background: Neoadjuvant chemotherapy (NACT) is an integral component for locally advanced and large operable breast cancer. The sequence of taxanes followed by anthracyclines has been the standard of care for almost 20 years. Eribulin (E) is a synthetic analogue of halichondrin B with distinct mechanism of action as microtubule dynamics inhibitor. The FDA approved E in 11/2010 for the treatment of patients (pts) with metastatic breast cancer who have previously received at least two chemotherapeutic regimens for the treatment of metastatic disease. Research Hypothesis: Sequential administration of eribulin followed by FAC/FEC-regimen, would have greater pathologic complete response (pCR) rate than sequential administration of paclitaxel followed by FAC/FEC-regimen as primary systemic therapy for woman with operable breast cancer.

Methods: This is a phase II, randomized, single institution, open label study. Pts were randomized 1:1 to receive E (1.4 mg/m2 d1 and d8 q 21 days x 4) or paclitaxel (P) (80 mg/m2 weekly x12). Both arms received FAC/FEC regimen x 4 doses followed by surgery. Eligible pts were women age 18 or older, Karnosfky PS 80 – 100, histologically confirmed invasive breast cancer, clinical T2-T3, N0-3, M0, HER2-negative. Baseline LVEF of &gt; 50% and normal hematology, liver and kidney laboratory function tests. Primary endpoint was pathologic complete response (pCR/RCB-0) assessed by residual cancer burden (RCB). [Symmans F, 2007]. This protocol (2012-0167) IRB of The University of Texas, MD Anderson Cancer Center.

Results: A preplanned interim analysis aimed to validate trial assumption was conducted after treatment of 54 randomized pts. Between 8/2012 to 7/2014, 54 pts were randomized and 49 were evaluable for pCR(27 P arm and 22 E arm). Tumor response by RCB is shown in the table. pCR rates were 30% and 4.5% in the P and E arm, respectively.

Table 1.ResponsePaclitaxel - FAC/FEC Arm (N=27)Eribulin - FAC/FEC Arm (N=22)RCB 0 (pCR)8 (30%)1 (4.5%)RCB I6 (22.2%)1 (4.5%)RCB II9 (33%)10 (45%)RCB III4 (14.8%)10 (45%)

53 pts were evaluable for toxicity. The combination was safe with mostly grade 1 and 2 toxicities in both arms. In the P arm grade 3 peripheral neuropathy and neutropenia was seen in 3% and 7%, respectively. In the E arm one patient died due to multiorgan failure during cycle 1. There was no other grade 3-5 toxicity. Biomarker analysis using CTCs by AdnaTest Breast were evaluated in 39 pts at baseline. 5/39 pts were positive for CTCs. 3 pts had transcripts for EpCAM, 1 for Muc-1 and another had both. 30 pts had an additional sample post therapy. 2 pts were positive for CTC at baseline as well as at follow up (FU) visit at 180 days. None of the samples showed CTC-EMT at baseline or at FU visits.

Conclusions: The interim analysis demonstrated that E arm lead to significantly lower pCR/RCB1 rate compared to P arm. Ongoing biomarker analyses include TIL, hot spot mutation analysis (HSMA) and molecular inversion probes (MIP) will be presented at the time of the meeting. Clinical trial information: NCT01593020.

Citation Format: Alvarez RH, Koenig KB, Ensor JE, Ibrahim NK, Chavez-MacGregor M, Litton JK, Schwartz Gomez JK, Cyriac A, Krishnamurty S, Caudle AS, Shaitelman SF, Whitman GJ, Booser DJ, Reuben JM, Valero V. A randomized phase II neoadjuvant (NACT) study of sequential eribulin followed by FAC/FEC-regimen compared to sequential paclitaxel followed by FAC/FEC-regimen in patients (pts) with operable breast cancer not overexpressing HER-2. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P1-14-04.

Development and usability testing of a web-based cancer symptom and quality-of-life support intervention

The feasibility and acceptability of computerized screening and patient-reported outcome measures have been demonstrated in the literature. However, patient-centered management of health information entails two challenges: gathering and presenting data using "patient-tailored" methods and supporting "patient-control" of health information. The design and development of many symptom and quality-of-life information systems have not included opportunities for systematically collecting and analyzing user input. As part of a larger clinical trial, the Electronic Self-Report Assessment for Cancer-II project, participatory design approaches were used to build and test new features and interfaces for patient/caregiver users. The research questions centered on patient/caregiver preferences with regard to the following: (a) content, (b) user interface needs, (c) patient-oriented summary, and (d) patient-controlled sharing of information with family, caregivers, and clinicians. Mixed methods were used with an emphasis on qualitative approaches; focus groups and individual usability tests were the primary research methods. Focus group data were content analyzed, while individual usability sessions were assessed with both qualitative and quantitative methods. We identified 12 key patient/caregiver preferences through focus groups with 6 participants. We implemented seven of these preferences during the iterative design process. We deferred development for some of the preferences due to resource constraints. During individual usability testing (n = 8), we were able to identify 65 usability issues ranging from minor user confusion to critical errors that blocked task completion. The participatory development model that we used led to features and design revisions that were patient centered. We are currently evaluating new approaches for the application interface and for future research pathways. We encourage other researchers to adopt user-centered design approaches when building patient-centered technologies.

Atypical and columnar cell lesions in breast needle biopsies for indeterminate microcalcifications: predictors of higher risk findings requiring surgical excision

Abstract #3004

Background: Columnar cell lesions (CCL) are frequently identified in biopsies performed for calcifications. The primary objective of this study is to identify histologic features of CCL in directional vacuum-assisted needle biopsies (DVAB) that could predict a higher risk lesion requiring excision.&#x2028; Design: From June 2002 to January 2008, we retrieved the materials for 88 DVABs and their subsequent segmental resections. We excluded carcinomas, intraductal papillomas, fibroadenomas, adenosis, or radial scars. Using World Health Organization criteria, CCLs were classified as CCL with no atypia (CCL-NA), CCL with cytologic atypia (CCL-CA; flat epithelial atypia), or architectural atypia/atypical ductal hyperplasia (CCL-ADH) in the background of CCL. In cases with atypia, the involved terminal duct lobular units/ducts (TDLUs) were quantified. Atypical cases were separated two groups as follows: minimal ADH (≤2 TDLU and minimal cytologic atypia) and extensive ADH/borderline DCIS (≥3 TDLU and/or significant cytologic atypia bordering on DCIS).Radiographs were evaluated to assess amount of calcifications removed. Biopsy results were correlated with the subsequent surgical excision results.&#x2028; Results: Eight of 88 biopsies were excluded because of overdiagnosis or because the lesions were not associated with microcalcifications. The remaining 80 biopsies included 5 cases of CCL-NA, 4 pure CCL-CA, 20 mixed CCL-CA + ADH and 51 pure CCL-ADH. Microcalcifications were identified in association with the targeted lesions. There were no higher risk findings in any of the resection specimens with CCL-NA and pure CCL-CA. Ten of the atypical lesions showed carcinoma on excision (9 DCIS and 1 invasive cancer). Twenty-five percent (9/36) extensive ADH/borderline DCIS showed carcinoma on excision. In contrast, only 2.5% (1/39) of minimal ADH cases showed carcinoma on excision. Radiologic review revealed residual abnormality in this one case.&#x2028; Conclusion: In summary, the extent of lobular involvement by atypia (CCL-CA or CCL-ADH) correlated with the presence or absence of higher risk lesions upon subsequent excision. These results suggest that additional surgery may not be necessary for CCL-CA or CCL-ADH that is limited to ≤ 2 foci provided sampling of the microcalcifications is adequate.

Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 3004.

Effectiveness of screening women at high risk for breast cancer with alternating mammography and MRI

Abstract #5015

Background&#x2028; MRI has been found to be more sensitive than mammography in the detection of breast cancer, with sensitivity of 71% to 100%, compared to 16 to 40% for mammography. Therefore, MRI has been introduced as an adjunct to annual screening mammography and clinical examination (CE) in women at high risk for developing breast cancer. We investigated the efficacy of alternating screening mammography with MRI every 6 months in women at high risk for developing breast cancer.&#x2028; Methods&#x2028; Retrospective chart review was performed on 334 consecutive women who were seen in the high risk clinic at one institution, from January 1997 to December 2007. Patients with increased risk for breast cancer and had an MRI were included in this study. Women with hereditary breast and ovarian cancer syndrome, personal breast cancer history, a biopsy of atypia or lobular carcinoma in situ (LCIS), or a 20% or higher lifetime risk of developing breast cancer by the Gail model were included. Mammography, ultrasound, and MRI findings were reviewed, along with biopsy results.&#x2028; Results&#x2028; 86 of 334 (26 %) patients underwent annual screening mammography and MRI, with mammography and MRI alternating every six months. The remaining 248 patients underwent prophylactic mastectomies or were treated with chemoprevention. CE was performed every six months. 47% of 86 patients completed the 1st round of MRI surveillance, 32% completed the 2nd round, 15% completed the 3rd round, and 6% completed the 4th round. Of 86 patients who underwent MRI screening, 70 had BRCA mutations, and 16 had a history of breast cancer , a biopsy of atypia or LCIS, or a 20% or higher lifetime risk of developing breast cancer. The median follow-up period was 2 years (range, 1 to 4 years).&#x2028; Nine cancers (6 invasive ductal, 1 invasive lobular, and 2 ductal carcinomas in situ) were detected in 7 of 86 (8.1%) women who underwent screening MRI; two women had bilateral cancers. Among the 9 cancers, 4 (44 %) were identified by MRI but not by mammography, 4 (44 %) were identified by both MRI and mammography, and 1 (11 %) was not identified by mammography or MRI. No cancer was seen by mammography only. Of the 9 tumors, 8 cancers were detected by MRI (sensitivity of 89%). One cancer was identified by mastectomy only (1 mm focus of DCIS). In 4 of the 8 MRI detected-cancers, the mammogram performed six months earlier was normal or demonstrated benign findings. The mean size of the cancers was 10.4 mm (range, 1 to 25 mm). At MRI, 5 cancers appeared as a mass and 3 cancers demonstrated non-mass like enhancement. Metastatic ipsilateral axillary lymphadenopathy was seen in association with 3 of the 9 cancers detected.&#x2028; Conclusions&#x2028; In women at high risk for developing breast cancer, alternating mammography and MRI at six month intervals demonstrated that no cancers were detected by mammography only. Future prospective studies should be performed to evaluate screening with annual mammography and MRI, alternating at six month intervals, versus annual MRI alone.

Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 5015.

The reliability of measuring physical characteristics of spiculated masses on mammography

The goal of this study was to assess the reliability of measurements of the physical characteristics of spiculated masses on mammography. The images used in this study were obtained from the Digital Database for Screening Mammography. Two experienced radiologists measured the properties of 21 images of spiculated masses. The length and width of all spicules and the major axis of the mass were measured. In addition, the observers counted the total number of spicules. Interobserver and intraobserver variability were evaluated using a hypothesis test for equivalence, the intraclass correlation coefficient (ICC) and Bland-Altman statistics. For an equivalence level of 30% of the mean of the senior radiologist's measurement, equivalence was achieved for the measurements of average spicule length (p<0.01), average spicule width (p = 0.03), the length of the major axis (p<0.01) and for the count of the number of spicules (p<0.01). Similarly, with the ICC analysis technique "excellent" inter-rater agreement was observed for the measurements of average spicule length (ICC = 0.770), the length of the major axis (ICC = 0.801) and for the count of the number of spicules (ICC = 0.780). "Fair to good" agreement was observed for the average spicule width (ICC = 0.561). Equivalence was also demonstrated for intraobserver measurements. Physical properties of spiculated masses can be measured reliably on mammography. The interobserver and intraobserver variability for this task is comparable with that reported for other measurements made on medical images.

Gene expression profiles predict complete pathologic response to neoadjuvant paclitaxel and fluorouracil, doxorubicin, and cyclophosphamide chemotherapy in breast cancer

PURPOSE

The goal of this study was to examine the feasibility of developing a multigene predictor of pathologic complete response (pCR) to sequential weekly paclitaxel and fluorouracil + doxorubicin + cyclophosphamide (T/FAC) neoadjuvant chemotherapy regimen for breast cancer.

PATIENTS AND METHODS

All patients underwent one-time pretreatment fine-needle aspiration to obtain RNA from the cancer for transcriptional profiling using cDNA arrays containing 30721 human sequence clones. Analysis was performed after profiling, and 42 patients' clinical results were available, 24 of which were used for predictive marker discovery; 18 patients' results were used as an independent validation set.

RESULTS

Thirty-one percent of patients had pCR (six discovery and seven validation), defined as disappearance of all invasive cancer in the breast after completion of chemotherapy. We could identify no single marker that was sufficiently associated with pCR to be used as an individual predictor. A multigene model with 74 markers (P <or=.09) was built using data from the discovery samples and tested on the validation samples. Overall, a 78% (14 of 18) predictive accuracy was observed, with a 100% (three of three) positive predictive value for pCR, a 73% (11 of 15) negative predictive value, a sensitivity of 43% (three of seven), and a specificity of 100% (11 of 11). The expected response rate to T/FAC neoadjuvant therapy in unselected patients is 28%.

CONCLUSION

Our results suggest that transcriptional profiling has the potential to identify a gene expression pattern in breast cancer that may lead to clinically useful predictors of pCR to T/FAC neoadjuvant therapy.

A case-control study of unilateral and bilateral breast carcinoma patients

BACKGROUND

Women with unilateral breast carcinoma are at increased risk for developing contralateral disease. The clinical significance of bilateral breast carcinoma has not been fully defined, and the subset of patients who may benefit from medical or surgical risk-reduction intervention has not yet been characterized. The purpose of this study was to evaluate risk factors and outcomes for bilateral breast carcinoma.

METHODS

A subject group of 70 bilateral breast carcinoma patients (62% metachronous) was matched by age and survival interval with a control group of 70 unilateral breast carcinoma patients. Median follow-up was 103 months.

RESULTS

Eighty-two percent of the unilateral patients and 80% of the bilateral patients had Stage I or II disease at diagnosis. Median age at presentation was 53 years. In the bilateral group, the contralateral cancer was diagnosed at the same or earlier stage than the first cancer in 87% of cases. Bilateral patients were significantly more likely to have multicentric disease and to have a positive family history for breast carcinoma compared with the unilateral group. There were no significant differences regarding history of exogenous hormone exposure, lobular histology, hormone-receptor status, or HER-2/neu expression. Five-year disease-free survival was 94% for the unilateral breast carcinoma patients and 91% for the bilateral breast carcinoma patients (P = 0.16).

CONCLUSIONS

Survival for patients with bilateral breast carcinoma is similar to that of patients with unilateral disease; however, prophylactic risk-reduction intervention for the contralateral breast should be considered in patients who have multicentric unilateral disease or a positive family history for breast carcinoma.