Background
It is estimated that in 2015, there will be 21,290 new cases of ovarian cancer diagnosed in the United States [
1]. While this type of cancer is rare compared with other types, the percentage of patients surviving 5 years after diagnosis is only 45.6 % [
1].
The current standard of care for late-stage ovarian cancer is cytoreductive surgery followed by 6–8 cycles of combination chemotherapy with a platinum-containing agent such as carboplatin [
2]. Patients who respond to platinum-based therapy and experience a relapse of ovarian cancer greater than 6 months after treatment completion are considered to have platinum-sensitive ovarian cancer [
3]. Those who experience a relapse during treatment or within 6 months after treatment are considered to have platinum-resistant ovarian cancer [
4] and are unlikely to respond to additional platinum-based treatments. However, the majority of patients at first recurrence of ovarian cancer have platinum-sensitive disease [
3] and standard therapy in these patients consists of retreatment with a platinum-containing regimen [
2]. It has also been found that deleterious mutations in
BRCA (breast cancer associated) genes are more prevalent in patients with platinum-sensitive ovarian cancer compared with platinum-resistant disease [
5]; this finding has implications for improving the treatment of recurrent ovarian cancer.
BRCA1 and
BRCA2 are genes that are critical for DNA repair through homologous recombination [
6]. Germline and somatic mutations in either
BRCA1 or
BRCA2 result in a significant increase in genomic instability and errors leading to carcinogenesis due to homologous recombination repair deficiency [
7]. Loss of function in both genes results in cell death, whereas loss of function in one gene allows for survival of cells with faulty DNA repair mechanisms [
8]. Therefore, patients with germline
BRCA1/2 mutations (
BRCAm) have enhanced susceptibility to agents that target DNA, such as platinum-containing agents, because of induction of synthetic lethality [
5,
9‐
11].
Owing to the enhanced susceptibility of platinum-containing agents in BRCAm platinum-sensitive recurrent ovarian cancer and the prevalence of BRCAm in recurrent disease, it is important to evaluate BRCA testing, treatment patterns and survival in patients who may benefit most from platinum therapy. The purpose of this study was to assess BRCA testing patterns, treatment patterns, and survival in patients with platinum-sensitive recurrent ovarian cancer.
Methods
Study design and data source
This was an observational cohort study of women with platinum-sensitive recurrent (PSR) ovarian cancer treated at the Huntsman Cancer Institute (HCI) in Salt Lake City, Utah. The Huntsman Cancer Institute tumor registry (HCI-TR) was used to identify patients with site and histology codes for epithelial ovarian cancer between January 1, 1995 and December 31, 2012. Identified patients were linked to electronic health record data from the University of Utah Enterprise Data Warehouse (EDW), which provided patient demographic and clinical information, including laboratory test results, medications, procedures, health status and physician notes.
Study population
Included patients were 18 years and older, with a diagnosis of epithelial ovarian cancer (ICD-10/ICD-O C56.9), fallopian tube cancer (C57.0), or primary peritoneal cancer (C48.1–48.3) in the HCI-TR, and had at least two health care visits separated by ≥30 days with ICD-9 codes for ovarian (183.X), fallopian tube (183.2) or primary peritoneal cancer (158.x) at HCI. Patients who received a platinum-based regimen (carboplatin or cisplatin) for initial systemic treatment of ovarian, fallopian tube, or primary peritoneal cancer and had a platinum-free interval (PFI) of at least 6 months before detection of recurrence were identified in the EDW or via chart review, classified as PSR ovarian cancer patients and included in the final analysis. Patients were excluded if they had a diagnosis of in situ ovarian cancer, had no documented use of platinum-containing agents, were platinum refractory, had a PFI of <6 months, or had no evidence of recurrence. The date of the first recurrence in the EDW was considered the index date for time to event data. The HCI-TR provided the date, stage, pathology, histological grade, and primary tumor site at ovarian cancer diagnosis. Clinical characteristics such as family history of hereditary breast or ovarian cancer and personal history of breast cancer were assessed at recurrence using chart review in the EDW. The University of Utah’s Institutional Review Board and the HCI Clinical Cancer Investigations Committee approved this study.
Classification of platinum-sensitive recurrent ovarian cancer
All patients who received platinum-based first-line treatment were identified in the EDW or via chart review and assessed for response and the PFI. PFI was defined as the number of months from the last platinum dose to recurrence. Those without evidence of recurrence after first-line treatment were excluded. Patients relapsing during first-line treatment or within 6 months of the last platinum treatment were categorized as platinum refractory and also excluded. Platinum sensitivity was defined as a response to first-line platinum treatment and a PFI of ≥6 months or physician-documented platinum sensitivity.
BRCA testing and status
BRCA status was ascertained through chart review. All medical records with mention of BRCA were reviewed by a semi-automated keyword search of the electronic notes with a custom text search tool that searched keywords or patterns of words using Boolean constructs. Patients were classified as BRCAm (BRCA1/2 positive) if a deleterious BRCA1 or BRCA2 germline mutation was documented in the medical record. Patients were classified as BRCAwt (BRCA wild type or variant of uncertain significance) if BRCA testing was conducted and documentation of the wild-type BRCA gene (no deleterious mutation detected) or a variant of uncertain significance was recorded. Lastly, patients were classified as untested if BRCA testing was not performed or if BRCA status was not recorded in the medical record.
Treatment patterns
Primary treatment modalities at the time of ovarian cancer diagnosis were categorized as systemic chemotherapy, radiation, or cytoreductive surgery; these treatments were also evaluated from recurrence until death or last follow-up. Systemic chemotherapy was further categorized as first-line, second-line, and third-line treatment based on the treatments received from recurrence as assessed by chart review. Systemic treatment lines after the initial treatment at recurrence were defined as a change in systemic treatment (addition and/or deletion of drug) due to disease progression, adverse events, or tolerability. Retreatment with the same systemic treatment after a delay in treatment as a result of an adverse event or the inability to tolerate the regimen was not considered a new treatment line. In patients who had a complete response to treatment and in whom systemic treatment was subsequently discontinued, an additional treatment line was considered if the same or alternative systemic treatment was restarted at subsequent disease recurrence/progression.
Statistical analysis
Descriptive statistical analyses, including mean and standard deviation for continuous variables and count and percentage for categorical variables, were performed. Student t-test and Wilcoxon rank-sum test were used for continuous variables and Fisher’s exact test was used for categorical variables. A significance level of 5 % was utilized for this study.
Logistic regression was conducted to estimate the likelihood of patients receiving BRCA testing (BRCAm or BRCAwt) versus untested patients. The covariates in the model included age at ovarian cancer diagnosis, ethnicity, family history of breast or ovarian cancer, personal history of breast cancer, ovarian cancer diagnosis stage, pathology, primary tumor site, and year of ovarian cancer diagnosis.
Kaplan-Meier methodology was used to evaluate survival from index date (date of PSR) until death or last follow-up. Survival was stratified by BRCA status and compared by log-rank test. A Cox proportional hazard model was constructed from recurrence to death or time of last follow-up, whichever occurred earlier, by including patient demographics, clinical characteristics, family history or personal history of breast or ovarian cancer, year of ovarian cancer diagnosis, and BRCA testing status (tested vs. untested) as covariates in a single model run.
Discussion
This study provides a unique perspective into BRCA testing, clinical characteristics, treatment patterns, and survival outcomes in unselected, consecutive patients with PSR ovarian cancer. These data suggest that tested patients are younger at diagnosis, receive more treatment lines and have improved survival compared with untested patients.
This study demonstrates that in an academic oncology center with extensive genetic services support, approximately 24 % (
n = 40) of patients with PSR ovarian cancer (
n = 168) were tested for
BRCAm. Of those who were tested, 37.5 % (
n = 15) tested positive for a deleterious mutation in
BRCA1 or
BRCA2, indicating a high pre-test probability for
BRCAm, since it is thought that only 12–15 % of invasive ovarian cancers are associated with
BRCAm [
12,
13]. The high probability for
BRCAm was also supported by other factors such as family history of breast or ovarian cancer, which was observed in all
BRCAm patients and is known to be positively associated with
BRCA testing.
BRCA1/2 testing was made commercially available in 1996 by Myriad Laboratories, Inc. Utilization of
BRCA testing increased during the study period, with more patients with PSR ovarian cancer being
BRCA tested after 2006, which is substantiated by outside reports of increased utilization of
BRCA testing [
14]. Also, it was in 2006 that HCI instituted a Hereditary Risk Evaluation Program, which may have contributed to increased
BRCA testing in subsequent years and may account for the variation in the number of tested patients before and after 2006.
These data affirm that patients with a family history of hereditary breast or ovarian cancer are more likely to have been BRCA tested, with all BRCAm patients having a history compared with 44 % of BRCAwt and 20 % of untested patients. Additionally, BRCAm patients (33 %, n = 5) were more likely to have a personal history of breast cancer, either prior to ovarian cancer diagnosis or concurrently, compared to untested (5 %, n = 7) and BRCAwt patients (20 %, n = 3).
BRCA-tested patients were also significantly younger than untested patients. The median age for
BRCAm patients was similar to those in other reports [
15]. The similar age of
BRCAm and
BRCAwt patients potentially represents a channeling bias as patients diagnosed at a younger age are more likely have a
BRCAm and clinicians would be more likely to offer genetic testing to younger patients based on guideline recommendations [
2].
Overall, younger age at diagnosis of ovarian cancer, family history of hereditary breast or ovarian cancer, and personal history of breast cancer were all significant predictors of
BRCA testing in this population and suggests that testing was motivated by genetic risk assessment. Additionally, based on other population-based studies indicating a
BRCAm rate of 12–15 % [
12,
13] in invasive ovarian cancer, our cohort would be expected to contain 20–25
BRCAm carriers. Therefore, over the study period, genetic risk assessment and
BRCA testing potentially discovered 60–75 % of the expected
BRCAm carriers. However, the expected
BRCAm rate may be higher in patients with PSR ovarian cancer. Universal
BRCA testing is now recommended for all patients with high-grade serous ovarian cancer and would have likely identified additional
BRCA carriers [
2].
Our study identified 11
BRCA1 and 4
BRCA2 mutation carriers thus a
BRCA1-to-
BRCA2 ratio of ~3:1. There are geographic and ethnic differences in
BRCA1 and
BRCA2 mutations [
16]. Common founder
BRCA1/2 mutations with a significant role in the population occur in Ashkenazi Jews, as well as in Iceland, Russia, Germany, Hungary, Norway, Finland, Sweden, Denmark, France, the Netherlands, and the UK [
16]. Overall, 8–40 % of all
BRCA1 mutations have been identified in families from the UK, USA, France, Germany, Italy, and the Netherlands [
16]. In 13 studies containing at least 60 families in which one or more cases of ovarian cancer were ascertained, the frequency of
BRCA1 ranged from 24.2 to 76.2 % and the frequency of
BRCA2 ranged from 1.0 to 16.7 % [
16]. The overall ratio of
BRCA1 to
BRCA2 mutations ranged from 2:1 to 62:1 [
16]. These 13 studies mostly took place in European countries (aside from Australia and the USA). Thus, the
BRCA1:
BRCA2 ratio of 3:1 in our study is consistent given that the majority of Utah inhabitants are of Northern European descent.
Improved ovarian cancer survival outcomes in
BRCAm carriers have been reported [
15,
17]. The largest and most recent meta-analysis demonstrated that
BRCA1 and
BRCA2 mutations have positive prognostic effects on ovarian cancer overall and progression-free survival [
18]. The results of this meta-analysis revealed that
BRCA1 mutation carriers were associated with better overall survival than non-carriers, with a pooled HR of 0.76 (95 % CI: 0.70, 0.83) [
18]. However,
BRCA2 mutation carriers were associated with even better survival outcomes compared with non-carriers, with a pooled HR of 0.58 (95 % CI: 0.50, 0.66) [
18]. Platinum-free survival was also improved in
BRCA1 mutation carriers (HR = 0.65; 95 % CI: 0.52, 0.81) and
BRCA2 mutation carriers (HR = 0.61; 95 % CI: 0.47, 0.80) [
18]. Though we did not evaluate survival in
BRCA1 and
BRCA2 patients because of sample size limitations, median survival from recurrence was not significantly different between
BRCAm and
BRCAwt patients in our study, whereas it was lower in untested patients.
These data indicated that tested patients had better OS than untested patients. However, this result should be interpreted with caution as the tested patients tended to be younger and their diagnoses were more recent and thus changes in treatment practice may be contributing. Furthermore, these data support the necessity for testing patients to ensure early diagnosis and optimization of the treatment options.
Potentially contributing to the improved survival, tested patients also received more systemic treatment lines after diagnosis of PSR ovarian cancer than untested patients. However there were no significant differences between BRCAm and BRCAwt patients, potentially reflecting the lack of available targeted therapies. The proportion of patients who received a platinum-containing regimen at any time after PSR diagnosis was 87 % (n = 13) in BRCAm, 76 % (n = 19) BRCAwt, and 67 % (n = 118) in untested patients. Utilization of secondary cytoreductive surgery and palliative radiation therapy was similar between BRCAm (46.7 %) and BRCAwt (36 %) patients; however, fewer untested patients received palliative radiation therapy (19.5 %). The increased use of palliative radiation therapy in BRCA tested patients may be partially explained by their increased duration of survival and therefore increased opportunity to receive radiation therapy.
There are several limitations of this research that should be considered when interpreting the results of this study. These results were from a single institution and may not be generalizable to a larger population or geographic area where clinical practices differ. Furthermore, the demographic composition of the study resulted in under-representation of some demographic categories and small sample sizes for some groups, such as races other than Caucasian/white, precluding meaningful comparisons across some categories of interest. The practice patterns observed at the HCI, a National Cancer Institute Designated Center and member of the National Comprehensive Cancer Network, may not be typical of those in other types of treatment facilities and institutions.
Data from medical charts and tumor registries can be subject to missing data and coding errors. Although most of the demographic and clinical characteristics and treatment patterns of interest for this study were reported as known, there were several variables in the overall population with non-negligible missing or unknown values such as race/ethnicity, stage at ovarian cancer diagnosis, and ECOG performance status, which may have an impact on survival.
Lastly, this study was limited by the small number of BRCA tested patients (n = 40) versus the larger cohort of untested patients (n = 128) and in particular, the resulting small sample size of BRCAm carriers (n = 15) which did not allow for meaningful characterization of survival and treatment patterns between BRCA1 and BRCA2 mutation carriers. An inherent problem of observational studies is the possibility of selection bias; as this was an observational study of real-world BRCA testing practices, the BRCA test selection bias (younger age and more recent year of diagnosis) made it difficult to compare differences between BRCAm and BRCAwt patients. Also, owing to the limitations of available observational data, the untested group was used as a comparison group; this group was assumed to contain predominantly non-BRCAm carriers.
Competing interest
Sudhir K. Unni, Marisa B. Schauerhamer, Rishi Deka, Vanessa Stevens, Diana Brixner, and David Stenehjem were paid from a research grant provided for this study by AstraZeneca. Jerzy E. Tyczynski and Ancilla W. Fernandes are employees of AstraZeneca.
Authors’ contributions
SU helped in designing the study and conducting the statistical analysis. MS and RD performed the statistical analysis and helped to draft the manuscript. JT, AF, VS, and DB helped in conceptualizing and designing. DS was involved in conceptualizing, designing, analyzing, and interpreting the results. All authors read and approved the final manuscript.