The association between germline BRCA2 variants and sensitivity to platinum-based chemotherapy among men with metastatic prostate cancer
See editorial on pages 3441-4, this issue.
We acknowledge Dr. David Livingston for the kindly providing the plasmid pcDNA3-HA-BRCA2.
Abstract
BACKGROUND
Breast cancer 2 (BRCA2)-associated breast and ovarian cancers are sensitive to platinum-based chemotherapy. It is unknown whether BRCA2-associated prostate cancer responds favorably to such treatment.
METHODS
A retrospective analysis of a single-institution cohort of men with castration-resistant, metastatic prostate cancer was performed to determine the association between carrier status of pathogenic BRCA2 germline variants and prostate-specific antigen response to carboplatin-based chemotherapy. From 2001 through 2015, 8081 adult men with prostate cancer who had a consultation and/or underwent treatment at Dana-Farber Cancer Institute provided blood samples and consented to analyses of biologic material and clinical records. A subgroup of 141 men received at least 2 doses of carboplatin and docetaxel for castration-resistant disease (94% were also taxane refractory). These patients were categorized according to the absence or presence of pathogenic germline mutations in BRCA2 based on DNA sequencing from whole blood. The primary outcome was the response rate to carboplatin/docetaxel chemotherapy, defined according to a decline in prostate-specific antigen that exceeded 50% within 12 weeks of initiating this regimen. Associations between BRCA2 mutation status and response to carboplatin-based chemotherapy were tested using the Fisher exact test, with a 2-sided P value < .05 as the threshold for significance.
RESULTS
Pathogenic germline BRCA2 variants were observed in 8 of 141 men (5.7%; 95% confidence interval, 2.5%-10.9%). Six of 8 BRCA2 carriers (75%) experienced prostate-specific antigen declines >50% within 12 weeks, compared with 23 of 133 noncarriers (17%; absolute difference, 58%; 95% confidence interval, 27%-88%; P < .001). Prostate cancer cell lines functionally corroborated these clinical findings.
CONCLUSIONS
BRCA2-associated, castration-resistant prostate cancer is associated with a higher likelihood of response to carboplatin-based chemotherapy than non-BRCA2–associated prostate cancer. Cancer 2017;123:3532-9. © 2017 American Cancer Society.
INTRODUCTION
Breast and ovarian cancer families who have segregating deleterious breast cancer 2 (BRCA2) variants have an increased risk of developing prostate cancer (PCa).1-4 Multiple studies—several based on cases ascertained through breast and ovarian cancer pedigrees, early age of onset, and/or Ashkenazi Jewish or Icelandic ancestry—have reported associations between BRCA2 variants and features of aggressive PCa, including higher grade, more advanced stage at diagnosis, and inferior overall survival.4-10
Emerging evidence suggests that status of the DNA damage-repair pathway influences sensitivity to platinum-based chemotherapy. Patients with mutations in BRCA1 or BRCA2 create a defect in homologous DNA recombination, making cells that harbor these mutations susceptible to DNA-damaging agents, such as cisplatin or carboplatin. BRCA1-associated and BRCA2-associated breast and ovarian cancers, for example, are more responsive to platinum-based chemotherapy than non-BRCA2–mutated cancers.11, 12 Therefore, it has been hypothesized that patients who have PCa with DNA repair defects (in the germline or somatic genomes) would similarly have a higher response rate to platinum-based chemotherapy than noncarriers.13-15
However, platinum agents are not used routinely for the treatment of PCa. Although case series and early phase studies of platinum-based chemotherapy suggested activity in some patients with metastatic, castration-resistant PCa (mCRPC),14, 16 a phase 3 trial of 1 agent, satraplatin, failed to demonstrate a survival benefit in a general mCRPC population.17 Consequently, platinum agents are not currently included in clinical practice guidelines, except for the rare small cell variant of PCa.18 In the current study, we hypothesized that men with metastatic PCa harboring protein-truncating germline variants in BRCA2 might respond more favorably to platinum-based chemotherapy compared with otherwise similar men lacking these variants.
MATERIALS AND METHODS
Study Cohorts
The Gelb Center cohort at Dana-Farber Cancer Institute (DFCI) has been previously described.19 From 2001 to 2015, 8081 men with PCa provided blood samples for DNA extraction and consented to an analysis of blood specimens and medical records on DFCI Institutional Review Board-approved protocol 01-045. We identified men who received at least 2 doses of carboplatin/docetaxel chemotherapy for mCRPC. Patients in the database who received only 1 cycle were excluded. Men with small cell and neuroendocrine histology (based on examination of primary prostate tissue in the majority of cases) were excluded. Docetaxel/carboplatin therapy had been tested at DFCI in a dual-institution, phase 2 study that accrued 34 patients between 2004 and 2006.16 Fourteen taxane-refractory patients from that trial were included in the current study along with an additional 130 men who received the regimen but were not enrolled in a therapeutic clinical trial. In total, 144 men met inclusion criteria, and all received at least 2 doses of carboplatin (area under the curve, 3-5) in combination with docetaxel (range, 60-75 mg/m2) every 3 weeks. All patients had castration-resistant disease, and 133 men were taxane resistant at the time they initiated docetaxel/carboplatin. Three men (all BRCA2 noncarriers) were excluded because their prostate-specific antigen (PSA) levels were not adequately ascertained. Neither family history nor germline genetic testing was a selection criterion for the receipt of this chemotherapy regimen.
Genetic Sequencing/Analysis
By using DNA isolated from blood samples, next-generation targeted sequencing was performed using the Illumina TruSight Cancer Sequencing Panel (Illumina, Inc., San Diego, Calif; http://www.illumina.com/products/trusight_cancer.html, Accessed March 9, 2016.). The panel includes 94 genes (35 have been identified as being involved in human DNA repair20). Sequencing libraries were created according to the manufacturer's instructions.
Variant Call Pipeline
To identify genomic variants, we used 2 independent, commonly used, standard variant calling pipelines (GATK21, 22 and samtools/bcftools23, 24). Basic statistical analyses were performed, including read numbers (average, 300,000) and alignment rates (range, 83.41%-89.78%).
Bowtie2 aligner was used (Johns Hopkins University, Baltimore, Md), and polymerase chain reaction duplicates were removed by samtools rmdup followed by variant calling using samtools mpileup (base quality, >30) and bcftools call. Generated Variant Call Format files were converted and annotated using ANNOVAR (ANNOtate VARiation) software.25 High-quality ANNOVAR scores (Q > 200) were called, and frameshift (insertion/deletion) and stop-gain mutations were collected. All candidate BRCA2 variants were validated by Sanger sequencing.
Six distinct BRCA2 variants distributed across 8 carriers met our criteria (indel resulting in frameshift or nonsense mutation). All variants included in the analysis were scored as pathogenic by the ClinVar database (accessed March 9, 2016)26 or according to guidelines set by the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.27 Variants in the other 35 DNA damage repair genes were subjected to the same criteria for inclusion.
Responsiveness to Platinum Chemotherapy
Platinum responsiveness was evaluated by comparing the baseline PSA obtained at initiation of carboplatin/docetaxel therapy with the PSA measurement obtained 12 weeks (±2 weeks) later. Men who survived for 12 weeks and had a PSA decline exceeding 50% were categorized as responders. Men who died within 12 weeks of treatment initiation or whose PSA levels did not decline by 50% were categorized as nonresponders. Twelve weeks was chosen to give patients ample time to experience a significant response while ensuring that variation in the number of cycles did not confound the analysis. Investigators without knowledge of participants' clinical characteristics or chemotherapy responses performed the sequencing and determined BRCA2 status. Clinical investigators ascertained PSA response without knowledge of BRCA2 status.
Statistical Methods
The prevalence of BRCA2 mutations was estimated with exact 95% confidence intervals (CIs). Clinical characteristics were compared according to BRCA2 carrier status using the Fisher exact test for categorical variables and t tests or nonparametric Wilcoxon rank-sum tests for continuous variables. Overall survival was defined from date of initiation of carboplatin/docetaxel through the date of death from any cause; patients who had not died were censored on the date they were last known to be alive. Associations of BRCA2 mutation status with PSA response and overall survival were tested using Fisher exact tests and log-rank tests. Adjusted odds ratios for the outcome of PSA response and hazard ratios for the outcome of overall survival were obtained, respectively, by logistic and Cox proportional hazards regression models controlling for age, pretreatment PSA (log-transformed), year of treatment, and the presence of metastatic disease at initial diagnosis. The association of PSA response and overall survival was evaluated with the log-rank test, limiting the analysis to 130 patients who survived to the landmark time of 12 weeks (ie, the time at which PSA response was ascertained). A 2-sided P value < .05 was considered statistically significant. Statistical analyses were performed using Stata version 13.1 (StataCorp, College Station, Tex).
Functional Analyses of BRCA2 Mutations
PC-3 and LNCaP cells were obtained from the American Type Culture Collection (Manassas, Va), maintained in RPMI-1640 supplemented with 10% fetal bovine serum, and seeded into 96-well plates (1 × 104 cells/well), followed by small interfering RNA (siRNA) and plasmid transfection. BRCA2 or nonspecific siRNA (final, 40 nM) was transfected using Lipofectamine RNAiMAX (Invitrogen, Carlsbad, Calif), and wild-type BRCA2 or vector control (50 ng/well) using X-tremeGENE HP reagent (Roche Diagnostics, New York, NY) according to the manufacturer's instructions. Two days after transfection, carboplatin or solvent was added, and the cells were cultured for 3 days before we ran AlarmaBlue assays (Invitrogen). The following siRNAs were used: control siRNA and siBRCA2-1 (products SIC001 and NM_000059 [ID:SASI_Hs01_00121792], respectively; Sigma Aldrich, St. Louis, Mo); siBRCA2-2 (AACAACAATTACGAACCAAAC; synthesized by Sigma Aldrich); and pcDNA3-HA-BRCA2 (provided by Dr. David Livingston28). Western blot and quantitative reverse transcriptase-polymerase chain reaction analyses were performed as previously described.29 Anti-BRCA2 antibody was purchased from Santa Cruz Biotechnology (catalog no. SC-28235; Santa Cruz, Calif). The primers for quantitative reverse transcriptase-polymerase chain reaction were: hBRCA1-F, TGTGAAGGCCCTTTCTTCTG; hBRCA1-R, TCCCATCTGTCTGGAGTTGA; hBRCA2-F, CAAATAGACGAAAGGGGCAA; and hBRCA2-R, CCAGACTTTCAGCCATCTTG. All experiments were performed in 3 replicates, and these sets of experiments were repeated at least 3 separate times.
RESULTS
The characteristics of all 141 men with mCRPC who were assessed for germline BRCA2 status are provided in Table 1. All but 8 patients (94%) have died of PCa. Of the 141 men, 8 harbored a pathogenic germline BRCA2 variant (5.7%; 95% CI, 2.5%-10.9%). Thirty-five other genes implicated in the DNA damage repair pathway were analyzed for deleterious germline variants,20 but no individual gene other than BRCA2 was represented more than once across the cohort (Supporting Table 1; see online supporting information). Carriers of deleterious BRCA2 variants were diagnosed at a younger age and had lower PSA levels at diagnosis compared with noncarriers. Carriers also more frequently reported family histories of prostate, breast, or ovarian cancer (Table 1).
| No. of Patients (%) or Median [Range] | ||||
|---|---|---|---|---|
| Characteristic | Total | BRCA2 Carriers | BRCA2 Noncarriers | Pa |
| No. of patients | 141 (100) | 8 (5.7) | 133 (94.3) | |
| Age at diagnosis, y | 59 [40-80] | 53 [40-62] | 60 [40-80] | .03 |
| PSA at diagnosis, ng/mL | 13 [1-4600] | 5 [4-166] | 14 [1-4600] | .10 |
| Gleason score | .81 | |||
| 6 | 9 (6.4) | 0 (0) | 9 (6.8) | |
| 7 | 31 (22) | 1 (12.5) | 30 (22.6) | |
| 8-10 | 88 (62.4) | 6 (75) | 82 (61.7) | |
| Unknown | 13 (9.2) | 1 (12.5) | 12 (9) | |
| Initial treatment | .99 | |||
| Radical prostatectomy for localized disease | 57 (40.4) | 3 (37.5) | 54 (40.6) | |
| Radiation therapy for localized disease | 58 (41.1) | 4 (50) | 54 (40.6) | |
| Systemic treatment for metastases at diagnosis | 26 (18.4) | 1 (12.5) | 25 (18.8) | |
| Known family history of prostate cancer | .65 | |||
| Yes | 26 (18.4) | 2 (25) | 24 (18) | |
| No | 112 (79.4) | 6 (75) | 106 (79.7) | |
| Unknown | 3 (2.1) | 0 (0) | 3 (2.3) | |
| Known family history of breast or ovarian cancer | .01 | |||
| Yes | 18 (12.8) | 4 (50) | 14 (10.5) | |
| No | 118 (83.7) | 4 (50) | 114 (85.7) | |
| Unknown | 5 (3.5) | 0 (0) | 5 (3.8) | |
| Known family history of BRCA | .009 | |||
| Yes | 3 (2.1) | 2 (25) | 1 (0.8) | |
| No | 135 (95.7) | 6 (75) | 129 (97) | |
| Unknown | 3 (2.1) | 0 (0) | 3 (2.3) | |
| Ancestry | .99 | |||
| European American | 123 (87.2) | 8 (100) | 115 (86.5) | |
| African American | 4 (2.8) | 0 (0) | 4 (3) | |
| Hispanic | 2 (1.4) | 0 (0) | 2 (1.5) | |
| Unknown | 12 (8.5) | 0 (0) | 12 (9) | |
| Interval between date of diagnosis and carboplatin start, y | 6.3 [0.5-20.7] | 4.5 [1.1-13.7] | 6.3 [0.5-20.7] | .52 |
| Extent of metastatic disease at start of carboplatin | .45 | |||
| Bone and/or lymph node only | 91 (64.5) | 4 (50) | 87 (65.4) | |
| Visceral metastases beyond bone/lymph nodes | 49 (34.8) | 4 (50) | 45 (33.8) | |
| Unknown | 1 (0.7) | 0 (0) | 1 (0.8) | |
| PSA at start of carboplatin, ng/mL | 170 [0-9145] | 49 [1-515] | 204 [0-9145] | .02 |
| Progression on docetaxel before docetaxel/carboplatin | 133 (94.3) | 7 (87.5) | 126 (94.7) | .38 |
| Duration of treatment, wk | 13 [3-114] | 15 [13-33] | 12 [3-114] | .26 |
| No. of carboplatin cycles received | 4.5 [2-24] | 5 [4-15] | 4 [2-24] | .20 |
- Abbreviations: BRCA2, breast cancer 2; PSA, prostate-specific antigen.
- a P values for differences in clinical characteristics by BRCA2 carrier status were determined using the Fisher exact test for categorical variables; nonparametric Wilcoxon rank-sum tests for PSA values, interval between diagnosis and carboplatin start date, duration of treatment, and number of cycles; and the t test for age at diagnosis. Patients with unknown values were excluded from statistical comparisons.
Figure 1 displays the percentage PSA decline for each patient from baseline to 12 weeks of follow-up. Notably, among the 8 BRCA2 carriers, 6 (75%) had a PSA declines >50% within 12 weeks of treatment initiation, whereas 23 of the 133 men (17%) who lacked a deleterious BRCA2 germline variant had PSA declines >50% (absolute difference, 58%; 95% CI, 27%-88%; P < .001). This association corresponded to an unadjusted odds ratio of 14.3 (95% CI, 2.7-75.6). Adjusted for age, pretreatment PSA (log-transformed), year of treatment, and the presence of metastatic disease at initial diagnosis, the adjusted odds ratio was 11.3 (95% CI, 2.0-63.3; P = .006). Among BRCA2 carriers, 2 of 8 (25%) experienced a 90% decline in PSA compared with 6 of 133 noncarriers (4.5%). In terms of 30% PSA declines, responders comprised 6 of 8 carriers (75%) and 43 of 133 noncarriers (32.3%).
For the entire group of 141 men, the median survival from initiation of carboplatin/docetaxel was 10.2 months; among those who died, survival ranged from 1.54 to 98.1 months. In the landmark time analysis, a PSA response >50% was associated with prolonged survival. Among 130 patients who survived for more 12 weeks after the initiation of carboplatin/docetaxel, the median survival was 16.4 months for those who had a PSA decline >50% and 9.4 months for those who had no response or a decline <50% (P = .01; log-rank test) (Supporting Fig. 1; see online supporting information). Overall survival from treatment initiation was longer among carriers of pathogenic BRCA2 germline variants. Median survival from the start of chemotherapy was 18.9 months for carriers and 9.5 months for noncarriers (P = .03; log-rank test) (Supporting Fig. 2; see online supporting information). This association corresponded to an unadjusted hazard ratio of 0.41 (95% CI, 0.18-0.93) and an adjusted hazard ratio of 0.52 (95% CI, 0.22-1.25; P = .14).
To functionally evaluate the role of BRCA2 in carboplatin sensitivity, PC-3 and LNCaP PCa cell lines were analyzed. At baseline, PC-3 cells express wild-type BRCA2 at a high level, whereas LNCaP cells negligibly express BRCA2 (Supporting Fig. 3; see online supporting information). Suppression of BRCA2 in PC-3 cells significantly reduced cell viability in the presence of carboplatin (Fig. 2). Consistent with this, overexpression of BRCA2 in LNCaP cells enhanced cell viability under the same conditions (Fig. 2).
DISCUSSION
We identified a cohort of 141 men with metastatic, castration-resistant and docetaxel-resistant PCa from case series who were unselected for family history, age, or the presence of germline mutations and had received carboplatin/docetaxel chemotherapy. We observed that the presence of pathogenic germline BRCA2 variants was associated with increased responsiveness to platinum-based chemotherapy on the basis of the PSA decline at 12 weeks.
Overall, 20% of men had a PSA decline >50% within 12 weeks of treatment initiation. This is slightly less than the 25.4% response rate observed in the treatment arm of the phase 3 trial of satraplatin, a platinum agent used to treat patients with mCRPC in the docetaxel-refractory setting.17 This rate is considerably less than that observed in a phase 3 trial studying cabazitaxel, also in the docetaxel-refractory setting. In that trial, 39.2% of those who received cabazitaxel demonstrated a PSA response >50%.30 However, in our current cohort, when focusing on the subset of patients who carried pathogenic germline BRCA2 variants, 75% demonstrated a >50% PSA decline. This response rate is appreciably higher than that observed in previous studies of men with docetaxel-refractory mCRPC, including the cabazitaxel trial. This correlations between genotype and responsiveness to platinum chemotherapy is consistent with the experience observed among BRCA2 carriers with breast and ovarian cancers.31
Although these results suggest a survival advantage for men with BRCA2-associated mCPRC who receive treatment with carboplatin/docetaxel, the size of this retrospective study does not allow us to definitively reach that conclusion. It is unclear whether the trend in improved survival stems from the role of BRCA2 as a prognostic factor for mCRPC or from its role as a predictive factor corresponding to carboplatin responsiveness. Previous evidence demonstrating that BRCA2 mutations are associated with more aggressive PCa,7 coupled with the superior PSA decline observed among BRCA2 carriers in the current study, suggests the latter: BRCA2 status predicts a clinically meaningful response to carboplatin-based chemotherapy, which, in turn, may be associated with improved outcomes compared with noncarriers.
- Localized, high-grade PCa: Notably, 7 of 8 BRCA2 carriers in our mCRPC cohort were initially diagnosed with localized disease through PSA screening and were treated with curative intent, suggesting that there is an opportunity to alter the clinical course in this population. For BRCA2 carriers, a regimen like carboplatin/docetaxel may prove effective as neoadjuvant or adjuvant treatment along with radiation and hormone therapy.
- Hormone-naive, metastatic PCa: Docetaxel has emerged as part of standard, upfront management of men with newly diagnosed PCa metastases.32 BRCA2 carriers who present with metastases may benefit from the addition of carboplatin to androgen-deprivation therapy and docetaxel.
- Chemotherapy-naive mCRPC: Although trials using platinum in castrate-resistant PCa have been equivocal, as described above,16, 17, 33 platinum agents could be evaluated as part of first-line treatment for mCRPC in this subset of patients.
Our current findings should be viewed in the context of several key limitations. The size of the cohort—in particular, the mutation carrier cohort—could lead to clinical imbalances, such as differences in median PSA (Table 1), which, in turn, influence response. Also, although men in the cohort had castration-resistant and docetaxel-resistant disease, they varied in the number of prior treatments received and the era in which they were treated (26 were started on carboplatin in 2010 or later, and 4 had received abiraterone or enzalutamide after carboplatin). In addition, the point within the natural history of disease at which carboplatin/docetaxel therapy was initiated varied widely. Radiographic response to therapy could not be applied to the analysis, because many men had no imaging performed immediately before or after treatment. Those who did undergo imaging had different modalities used to evaluate tumor burden (eg, computed tomography scan vs bone scan) and underwent follow-up studies at inconsistent intervals. Among the patients who had PSA declines >50% on docetaxel/carboplatin, 2 of 6 carriers and 9 of 23 noncarriers underwent a computed tomography or bone scan during treatment. One carrier and 5 noncarriers had objective soft-tissue responses.
Similar to carboplatin, poly(adenosine diphosphate [ADP]-ribose) polymerase (PARP) inhibition has demonstrated increased activity against the metastatic tumors of germline BRCA2 carriers.13, 15 In a recent phase 2 study, PSA responses were consistently observed among BRCA2 carriers who received PARP inhibitors: 3 of 3 germline carriers experienced PSA declines >50%.13 PARP inhibitors may be an alternative for platinum chemotherapy and may have more a favorable side-effect profile.13 However, the costs of sustained PARP inhibition are currently higher than the costs of 6 cycles of carboplatin. Our results strongly motivate future studies evaluating both carboplatin and PARP inhibitors or their use in combination for men who have BRCA2-associated PCa.
In addition to inherited variation, there are acquired mutations in DNA damage-repair pathway genes, most commonly within BRCA2.34 Individuals whose tumors have acquired somatic mutations in these genes may also respond favorably to platinum agents.14 It is likely that several of the favorable responses observed among noncarriers in the current cohort had tumors with somatic mutations in DNA damage-repair pathway genes. Also, although we excluded men who demonstrated neuroendocrine differentiation, histology was assessed using primary tumors in most patients. We presume that many cancers in the cohort underwent neuroendocrine differentiation by the time they were receiving treatment with carboplatin. This could influence response rates and confound the analysis. Efforts to obtain and interrogate metastatic tissue and/or circulating biomarkers, in addition to germline testing, should be prioritized and presumably would expand the number of carboplatin/PARP-inhibitor candidates.
This study illustrates an important challenge in precision medicine—drawing inferences from necessarily small groups of patients. First, rates of deleterious germline variants in BRCA2 are relatively rare among patients with PCa. In a previous series, exome sequencing of BRCA2 revealed mutations in only 1.4% of men who had biopsy-detected PCa.9 Second, opportunities to study men with PCa who received treatment with platinum are limited, because this agent is rarely used. The only approved chemotherapy agents for metastatic adenocarcinoma of the prostate are taxanes and the anthracycline mitoxantrone. Platinum chemotherapy generally is not considered for mCRPC, except for men who have small cell or neuroendocrine histologies, both of which were excluded from our study.
In summary, these data strongly motivate BRCA2 sequencing in men with high-grade PCa in larger prospective studies in various clinical settings. More broadly, our results demonstrate the power of interrogating prior data sets with new technologies and ideas to rapidly actualize the goals of precision medicine.
FUNDING SUPPORT
Rebecca and Nathan Milikowsky (M.P.); the H.L. Snyder Medical Foundation (M.L.F.)
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
AUTHOR CONTRIBUTIONS
Mark M. Pomerantz: Conception and design of the study, acquisition, analysis and interpretation of data, writing–initial draft, and writing–final revisions. Sandor Spisák: Analysis and interpretation of data, writing–initial draft, and writing–final revisions. Li Jia: Conception and design of the study and acquisition, analysis, and interpretation of data. Angel M. Cronin: Analysis and interpretation of data, writing–initial draft, and writing–final revisions. Istvan Csabai: Analysis and interpretation of data. Elisa Ledet: Acquisition of data. A. Oliver Sartor: Acquisition, analysis, and interpretation of data and approval of the final article. Irene Rainville: Analysis and interpretation of data. Edward P. O'Connor: Acquisition of data. Zachary T. Herbert: Acquisition, analysis, and interpretation of data. Zoltan Szállási: Analysis and interpretation of data and approval of the final article. William K. Oh: Conception and design of the study, acquisition of data, and approval of the final article. Philip W. Kantoff: Design of the study, writing–initial draft, writing–final revisions, and approval of the final article. Judy E. Garber: Design of the study, writing–initial draft, writing–final revisions, and approval of the final article. Deborah Schrag: Design of the study, writing–initial draft, writing–final revisions, and approval of the final article. Adam S. Kibel: Conception and design of the study; acquisition, analysis, and interpretation of data; writing–initial draft; and writing–final revisions. Matthew L. Freedman: Conception and design of the study; acquisition, analysis, and interpretation of data; writing–initial draft; and writing–final revisions.
