Patterns of Cancer Family History and Genetic Counseling Eligibility Among African Americans With Breast, Prostate, Lung, and Colorectal Cancers: A Detroit Research on Cancer Survivors Cohort Study

Abstract

BACKGROUND:

Family history (FH) remains one of the strongest risk factors for many common cancers and is used to determine cancer genetic counseling (CGC) eligibility, but the understanding of familial cancer patterns in African Americans is limited.

METHODS:

This study evaluated cancer FH among African Americans with invasive breast cancer, prostate cancer, lung cancer, or colorectal cancer (CRC) in the Detroit Research on Cancer Survivors (ROCS) cohort. Associations between participant cancer type, site-specific FH, and meeting national guidelines for CGC were evaluated via logistic regression. Cancer FH patterns were evaluating via hierarchical clustering.

RESULTS:

Among 1500 ROCS participants, 71% reported at least 1 first-degree relative or grandparent with cancer. FHs of breast cancer, CRC, lung cancer, and prostate cancer were most common among participants with the same diagnosis (odds ratio [OR] for breast cancer, 1.14; P < .001; OR for CRC, 1.08; P = .003; OR for lung cancer, 1.09; P = .008; OR for prostate cancer, 1.14; P < .001). Nearly half of the participants (47%) met national CGC guidelines, and 24.4% of these participants met CGC criteria on the basis of their cancer FH alone. FH was particularly important in determining CGC eligibility for participants with prostate cancer versus breast cancer (OR for FH vs personal history alone, 2.91; 95% confidence interval, 1.94–4.35; P < .001). In clustering analyses, breast and prostate cancer FH–defined clusters were common across all participants. Clustering of CRC and breast cancer FHs was also observed.

CONCLUSIONS:

ROCS participants reported high rates of cancer FH. The high rate of eligibility for CGC among ROCS participants supports the need for interventions to increase referrals and uptake of CGC among African Americans.

INTRODUCTION

Breast cancer, prostate cancer, colorectal cancer (CRC), and lung cancer account for nearly half of all cancers diagnosed in the United States in 2019 as well as nearly half of all cancer deaths.¹ Although the overall incidence of these cancer types has declined in the last decade, racial disparities persist. In particular, African Americans are at increased risk for prostate cancer, CRC, and lung cancer and experience worse survival with all 4 cancer types in comparison with non-Hispanic Whites.^2–5 Although previous research has suggested that racial disparities in cancer incidence can be attributed to population differences in environmental, social, and genetic risk factors, the relative contribution of these factors to racial cancer disparities remains largely unknown.^6,7 Family studies remain a powerful approach for understanding cancer etiology because familial clustering reflects both shared environmental and genetic components⁸; however, our understanding of familial patterns of cancer risk in African Americans is limited.

Family history (FH) remains one of the strongest predictors of risk for these common cancers. Having a first- or second-degree relative with breast or prostate cancer is associated with a 2- to 3-fold increased risk of the same cancer.⁹ Although this is likely partially explained by shared lifestyle factors and environment, approximately 10% to 15% of breast cancers, 20% of prostate cancers, and 15% to 35% of CRCs are thought to be hereditary.^9–11 Rare high-penetrance mutations account for only ~6% to 15% of familial breast cancer, prostate cancer, and CRC cases, whereas common genetic variation is estimated to account for another 15% to 60% of this heritability.^12–14 Although there is a strong environmental component for lung cancer risk, familial aggregation of lung cancer beyond shared risk from smoking has also been described, and nearly 50 common genetic variants have been identified in lung cancer genome-wide association studies.¹⁵ However, the majority of these studies evaluating both rare and common genetic variations as risk factors have been performed in largely non-Hispanic White and/or European populations.

A better understanding of the familial contribution to cancer among African Americans has clear etiologic and clinical implications. Cancer FH is an important component for determining whether a patient with cancer is eligible to receive cancer genetic counseling (CGC) and/or testing to identify an underlying hereditary cancer syndrome. African Americans are less likely to be referred for or receive CGC and testing.^16,17 For individuals with potentially familial cancers, the need to receive CGC is critical because those found to have a hereditary cancer syndrome are at increased risk for second primaries and new primary cancers. There are also clear clinical guidelines for most familial cancers in terms of both treatment and follow-up, including consideration of prophylactic surgery, chemoprevention, and increased screening.^18,19 A better understanding of the cancer FH patterns among African Americans could help to address racial disparities in incidence and adverse outcomes and provide insights into cancer etiology.

MATERIALS AND METHODS

Detroit Research on Cancer Survivors Participants

Detailed methods for the Detroit Research on Cancer Survivors (ROCS) cohort have been described previously.²⁰ Briefly, the Detroit ROCS cohort study enrolls African Americans diagnosed with lung cancer, breast cancer, prostate cancer, or CRC and a subset of their caregivers. The eligibility criteria are as follows: 1) self-identification as African American or Black; 2) an age at cancer diagnosis of 20 to 79 years; 3) a date of diagnosis of or after January 1, 2013; 4) a diagnosis of first primary invasive lung cancer, female breast cancer, prostate cancer, or CRC; 5) residence in Wayne, Oakland, or Macomb County, Michigan, at the time of diagnosis; and 6) being alive at study contact. Participants complete a baseline survey and annual follow-up with optional biospecimen donation. The study protocol, all questionnaires, and study documents were reviewed and approved by the institutional review board at Wayne State University. The analyses presented here include the first 1500 participants enrolled into the Detroit ROCS study.

Patient Data Collection

Cancer FH was collected at the baseline survey. We collected information on 1) participant family size; 2) whether any first-degree relatives or grandparents had ever been diagnosed with breast cancer, prostate cancer, lung cancer, CRC, kidney cancer, liver cancer, ovarian cancer, pancreatic cancer, melanoma, or other cancers; and 3) family members’ ages at diagnosis. No participants reported an FH of melanoma. Details on the types of cancers included in the “other” cancer category were not captured for these first 1500 participants. Clinical data and tumor characteristics for participants were extracted from the Metropolitan Detroit Cancer Surveillance System registry database. Mismatch repair deficiency data were obtained from the Metropolitan Detroit Cancer Surveillance System (n = 6) or electronic pathology reports (n = 132). Because the TNM stage is used as the gold standard for informing prognosis and directing therapy,²¹ mismatch repair deficiency testing is not always performed. Mismatch repair deficiency was considered positive if 1) microsatellite instability (MSI) testing was positive (MSI-low or MSI-high) or 2) immunohistochemistry demonstrated an absence of protein expression for at least 1 mismatch repair protein (MLH1, MSH2, MSH6, or PMS2). Mismatch repair deficiency is observed in nearly all patients with CRC who have hereditary nonpolyposis colorectal cancer (HNPCC) syndrome, which is caused by germline mutations in mismatch repair genes; however, epigenetic silencing of MLH1 and resulting MSI in the tumor are also observed in sporadic tumors, and mismatch repair deficiency is present in up to 15% of CRCs.²²

National Comprehensive Cancer Network Criteria for CGC Referral

Published National Comprehensive Cancer Network (NCCN) guidelines were adapted on the basis of data collected in the ROCS baseline survey to define criteria for CGC referral eligibility.^18,19,23 Close relatives were defined as first- or second-degree relatives. The CGC guidelines for breast cancer included 1) a personal diagnosis at an age ≤50 years, (2) a personal history (PH) of the triple-negative subtype and an age at diagnosis ≤60 years, 3) at least 1 close relative diagnosed with breast cancer at an age <50 years, 4) at least 1 close relative diagnosed with ovarian cancer, or 5) 3 or more close relatives diagnosed with breast cancer, prostate cancer, melanoma, CRC, or kidney cancer. Guidelines for prostate cancer included 1) a PH of high-risk or very high-risk prostate cancer, 2) a PH of regional or metastatic prostate cancer, or 3) a PH of less than high-risk prostate cancer with a suggestive FH (ie, a father or brother or multiple relatives with prostate cancer diagnosed at an age <60 years; at least 1 relative with breast cancer, ovarian cancer, or pancreatic cancer; or at least 1 relative with CRC, ovarian cancer, pancreatic cancer, or kidney cancer). Referral guidelines for CRC included 1) a personal CRC diagnosis at an age ≤50 years, 2) a PH of mismatch repair–deficient CRC, 3) at least 1 close relative diagnosed with HNPCC at an age <50 years, or 4) at least 2 close relatives diagnosed with an HNPCC cancer at any age.

Statistical Analysis

Analyses were performed with SAS (version 9.4; SAS, Cary, North Carolina) and R software (https://cran.r-project.org/). Distributions of sociodemographic and clinical/tumor characteristics as well as FH variables are presented as frequencies and percentages for categorical variables and as medians and ranges for continuous variables. Differences in overall and site-specific cancer FHs were tested with chi-square or Fisher exact tests as appropriate. Associations between dichotomous FH variables and participant cancer type were estimated via unadjusted logistic regression. Associations between participant cancer type and criteria used to meet NCCN guidelines for CGC referral were estimated via multinomial logistic regression. UpSet plots were generated with the UpSetR R package. Hierarchical clustering was performed via the hclust function in the cluster R package using the number of relatives (first-degree relatives or grandparents) diagnosed with breast cancer, CRC, kidney cancer, lung cancer, liver cancer, ovarian cancer, prostate cancer, pancreatic cancer, or other cancers. The number of clusters for each participant cancer type was determined with the NbClust R package. Dendograms and heatmaps were generated with the dendextend and gplots R packages.

RESULTS

A total of 1500 Detroit ROCS participants were included in this analysis (Table 1). The median age at diagnosis was 59 years, and patients with breast cancer and CRC were diagnosed at a slightly younger age (57.5 years) than patients with lung and prostate cancer (60.5 years). Among CRC and lung cancer participants, approximately half were female. Overall, 58% of survivors reported an educational level of college or higher, and 52% of participants had at least some private insurance. Median numbers of siblings and children did not vary by cancer site. The majority of breast and prostate cancer survivors were stage I/II at diagnosis, whereas the majority of CRC and lung cancer survivors were diagnosed at stage III/IV. The majority of breast cancers were luminal A (56%), and they were followed by the triple-negative subtype (18%; Table 2). Among the CRC survivors, the tumor location was equally distributed between proximal and distal. Among the 80 CRC participants for whom data were available, 10% had tumors that demonstrated mismatch repair deficiency. The majority of lung cancers were non–small cell (92%), whereas the majority of prostate cancers (60%) had a Gleason score of 7 and prostate-specific antigen levels ≤10 ng/mL at diagnosis.

TABLE 1.

Demographics and Clinical Characteristics of the 1500 African American Detroit ROCS Participants

	All Sites		Breast		CRC		Lung		Prostate
Characteristic	No.	%	No.	%	No.	%	No.	%	No.	%
Total	1500		674		138		174		514
Age at diagnosis
<50 y	288	19	224	33	31	22	11	6	22	4
50–59 y	500	33	183	27	51	37	66	38	200	39
60–69 y	534	36	195	29	42	30	71	41	226	44
≥70 y	178	12	72	11	14	10	26	15	66	13
Age at diagnosis, median (range), y	59 (27–79)		57 (27–79)		57.5 (27–78)		60 (41–79)		61 (40–79)
Sex
Male	648	43	–	–	65	47	69	40	514	100
Female	852	57	674	100	73	53	105	60	–	–
Educationa
Less than high school	191	13	66	10	22	16	39	22	64	13
High school/GED	426	29	147	22	39	28	66	38	174	34
Some college/2-y degree	542	36	278	41	47	34	49	28	168	33
4-y college degree	167	11	92	14	11	8	13	8	51	10
Graduate/professional degree	162	11	84	12	18	13	6	3	54	11
Insurancea
Medicare only	208	14	74	11	14	10	33	20	87	17
Medicare + private	352	24	156	23	22	16	41	24	133	26
Medicaid	379	26	169	25	47	34	53	31	110	22
Medicaid + Medicare	104	7	32	5	12	9	17	10	43	9
Private	423	29	231	35	40	29	23	14	129	25
No insurance	11	1	4	1	2	1	2	1	3	1
AJCC stage
I	433	29	288	43	25	18	45	26	75	15
II	624	42	263	39	30	22	22	13	309	60
III	258	17	87	13	46	33	52	30	73	14
IV	166	11	32	5	34	25	55	32	45	9
Unknown	19	1	4	1	3	2	0	0	12	2

Abbreviations: AJCC, American Joint Committee on Cancer; CRC, colorectal cancer; GED, General Education Diploma; ROCS, Research on Cancer Survivors.

^aSome values were not reported or were unknown for education (n = 12) and insurance (n = 23).

TABLE 2.

Cancer Site–Specific Tumor Characteristics of Detroit ROCS Participants

Characteristic	No.	%
Breast cancer (n = 674)
Subtype
ER+ or PR+, HER2+ (luminal B)	96	14
ER+ or PR+, HER2− (luminal A)	378	56
ER−/PR−, HER2+	49	7
Triple-negative	118	18
Unknown	33	5
Colorectal cancer (n = 138)
Mismatch repair deficiencya
Normal	72	52
Deficient	8	6
Unknown	58	42
Location
Distal	66	48
Proximal	72	52
Lung cancer (n = 174)
Histology
Non–small cell	160	92
Small cell	14	8
Prostate cancer (n = 514)
Gleason scoreb
6	102	20
3 + 4	194	38
4 + 3	112	22
≥8	95	18
Unknown	11	2
PSA at diagnosisc
≤10 ng/mL	338	71
>10–20 ng/mL	72	15
>20 ng/mL	64	14

Abbreviations: ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; PR, progesterone receptor; PSA, prostate-specific antigen; ROCS, Research on Cancer Survivors.

^aDefined by microsatellite instability or an absence of mismatch repair protein expression by immunohistochemistry.

^bA low grade is indicated by a Gleason score of 6 or lower, an intermediate grade is indicated by a Gleason score of 3 + 4 or 4 + 3, and a high grade is indicated by a Gleason score of 8 or higher.

^cA normal PSA level is ≤4.0 ng/mL.

We next examined self-reported cancer FH among these participants (Table 3). Approximately 66% of participants reported at least 1 first-degree relative with cancer, and 71% reported at least 1 first-degree relative or grandparent with cancer; this did not differ by participant cancer site. FHs of breast cancer, CRC, lung cancer, and prostate cancer were most common among participants with the same diagnosis (odds ratio [OR] for breast cancer, 1.14; 95% confidence interval [CI], 1.09–1.19; P < .001; OR for CRC, 1.08; 95% CI, 1.03–1.14; P = .003; OR for lung cancer, 1.09; 95% CI, 1.02–1.15; P = .008; OR for prostate cancer, 1.14; 95% CI, 1.09–1.18; P < .001; Supporting Table 1). Participants diagnosed with prostate cancer in comparison with other cancer sites were also less likely to report an FH of ovarian cancer (OR, 0.97; 95% CI, 0.95–0.99; P = .005), breast cancer (OR, 0.90; 95% CI, 0.86–0.94; P > .001), or lung cancer (OR, 0.95; 95% CI, 0.92–0.99; P = .025). No significant differences were observed for FHs of kidney, liver, or pancreatic cancers.

TABLE 3.

Self-Reported Family History of Cancer Stratified by the Cancer Sites of Detroit ROCS Participants

	All Sites, No. (%)	Breast, No. (%)	CRC, No. (%)	Lung, No. (%)	Prostate, No. (%)	Pa
First-degree relatives
Any cancer family history	996 (66)	447 (66)	93 (67)	114 (66)	342 (67)	.99
Site-specific cancer family history
Breast cancer	279 (19)	159 (24)	24 (17)	22 (13)	74 (14)	<.001
CRC	113 (8)	50 (7)	18 (13)	11 (6)	34 (7)	.071
Lung cancer	230 (15)	100 (15)	19 (14)	40 (23)	71 (14)	.027
Prostate cancer	255 (17)	99 (15)	15 (11)	14 (8)	127 (25)	<.001
Kidney cancer	21 (1)	9 (1)	2 (1)	5 (3)	5 (1)	.33
Liver cancer	35 (2)	13 (2)	1 (1)	3 (2)	18 (4)	.14
Ovarian cancer	44 (3)	22 (3)	6 (4)	8 (5)	8 (2)	.096
Pancreatic cancer	45 (3)	19 (3)	3 (2)	8 (5)	15 (3)	.59
First-degree relatives + grandparents
Any cancer family history	1060 (71)	483 (72)	97 (70)	120 (69)	360 (70)	.88
Site-specific cancer family history
Breast cancer	345 (23)	203 (30)	28 (20)	30 (17)	84 (16)	<.001
CRC	143 (10)	67 (10)	23 (17)	11 (6)	42 (8)	.010
Lung cancer	271 (18)	126 (19)	24 (17)	44 (25)	77 (15)	.022
Prostate cancer	292 (19)	116 (17)	18 (13)	15 (9)	143 (28)	<.001
Kidney cancer	28 (2)	13 (2)	3 (2)	6 (3)	6 (1)	.28
Liver cancer	41 (3)	18 (3)	2 (1)	3 (2)	18 (4)	.45
Ovarian cancer	62 (4)	35 (5)	8 (6)	8 (5)	11 (2)	.044
Pancreatic cancer	60 (4)	31 (5)	4 (3)	8 (5)	17 (3)	.60

Abbreviations: CRC, colorectal cancer; ROCS, Research on Cancer Survivors.

^aComparison between the presence of a family history of cancer and participants’ cancer sites (chi-square test/Fisher exact test).

We were also interested in describing more complex patterns of FH to provide insight into shared etiologies from exogenous risk factors or underlying hereditary cancer syndromes. We first evaluated whether participants with breast cancer, prostate cancer, or CRC met NCCN guidelines for CGC referral. When we considered PH and FH together, nearly half of all Detroit ROCS participants (47%) met NCCN criteria for CGC referral (Fig. 1). Nearly a quarter (24.4%) met criteria for CGC referral on the basis of cancer FH alone. The relative contributions of cancer FH and PH to CGC eligibility were similar for participants with CRC in comparison with participants with breast cancer (OR for FH vs PH, 1.30; 95% CI, 0.64–2.66; P = .47). In contrast, prostate cancer participants were approximately 3 times more likely to be eligible for CGC on the basis of either FH alone or FH and PH combined than breast cancer participants (OR for FH vs PH, 2.91; 95% CI, 1.94–4.35; P < .001; OR for FH plus PH vs PH alone, 3.34; 95% CI, 2.05–1.94; P < .001).

Figure 1.

Proportions of participants who did not and did meet NCCN guidelines for CGC referral are shown by participant cancer site. The right-hand bars are shaded according to the factors that led to participants’ eligibility for CGC referral. CGC indicates cancer genetic counseling; NCCN, National Comprehensive Cancer Network.

We next examined the distribution of individual criteria used to classify participants as eligible for CGC. The majority of breast cancer participants were eligible on the basis of their age at diagnosis alone (50.3%) or in combination with other PH or FH factors (22.4%; Fig. 2A). Among the 17.5% of breast cancer participants eligible for CGC on the basis of FH alone, the vast majority (90.7%) had only 1 defining FH feature. For CRC participants, there was virtually no overlap between the 4 individual criteria (Fig. 2B). A young age at diagnosis was the most commonly fulfilled criterion (62%). For prostate cancer participants, an FH of breast cancer, CRC, ovarian cancer, pancreatic cancer, or kidney cancer was highly prevalent either alone (31.7%) or in combination with other PH or FH characteristics (21.8%; Fig. 2C). An FH of prostate cancer diagnosed at an age <60 years was rare (1.1%). For the 46.1% of prostate cancer participants who were eligible for CGC on the basis of PH alone, having a later stage prostate cancer or a Gleason score ≥8 was most common (~75% for each characteristic).

Figure 2.

Independent and joint distributions of PH and FH characteristics of ROCS participants eligible for CGC are shown via UpSet plots. Separate panels are provided for (A) 308 participants with BrCa, (B) 50 participants with CRC, and (C) 271 participants with PrCa. Each bar of the histograms represents the number of participants who were eligible for CGC according to the criteria listed in the matrix below, as indicated by the filled dots. Bars are labeled with the corresponding percentages among participants eligible for CGC within the participant cancer type. Filled turquoise dots correspond to FH criteria, and filled dark gray dots correspond to PH criteria. Vertical black lines that connect 2 or more dots indicate that participants met multiple criteria. BrCa indicates breast cancer; CGC, cancer genetic counseling; CRC, colorectal cancer; dx, diagnosis; FH, family history; HNPCC, hereditary nonpolyposis colorectal cancer; KidCa, kidney cancer; MMR, mismatch repair; OvCa, ovarian cancer; PanCa, pancreatic cancer; PH, personal history; PrCa, prostate cancer; PSA, prostate-specific antigen; ROCS, Research on Cancer Survivors; TNBC, triple-negative breast cancer.

Finally, we performed a hierarchical clustering analysis to agnostically capture patterns of cancer FH (Supporting Fig. 1). Across all 4 participant cancer types, we observed a single cluster defined by very low rates of cancer FH (Fig. 3). All cancer types also had clusters defined predominantly by very high rates of breast cancer FH or prostate cancer FH. Similarly, all participant cancer types had clusters defined predominantly by high rates of “other” cancer types, which included any cancer not explicitly asked about in the questionnaire. Only breast cancer and CRC participants had clusters defined primarily by CRC FH. CRC participants did not have a lung cancer FH–driven cluster in contrast to breast, prostate, and lung cancer participants. Among breast cancer participants, the prostate cancer and CRC FH clusters were also defined by ~20% of participants reporting breast cancer FH. Among participants with prostate cancer, there were 3 distinct prostate cancer FH–driven clusters: 1 with primarily a prostate cancer FH; 1 with ~15% to 20% of participants reporting breast cancer, CRC, lung cancer, liver cancer, and other cancer FHs; and 1 defined by high rates of both prostate and “other” cancer FHs. Among lung cancer participants, there were 2 lung cancer FH–dominant clusters. The first shared higher rates of liver, pancreatic, and “other” cancer FHs, whereas ~15% to 20% of participants reported FHs of breast and CRC in the second cluster.

Figure 3.

Site-specific cancer family history patterns are shown by cluster on the basis of hierarchical clustering results. Each dot represents the proportion of ROCS participants reporting a family history of cancer at a particular site (exact proportions are listed in Supporting Table 2). The size of the dots is directly proportional to that proportion, and a scale for interpreting dot sizes is included (see the gray dots in the upper right of the figure). Dot colors are provided only to aid in the visualization of family history patterns within each cluster. Each column of dots represents the relative contributions of cancer family histories at each of the sites indicated on the y-axis. For example, breast cancer participant cluster 2 is represented in the second column with orange dots; we can see that most participants in this cluster (~100%) reported a family history of breast cancer, a small proportion reported a family history of lung or other cancers (~15%–20%), and negligible proportions reported family histories of CRC, kidney cancer, liver cancer, ovarian cancer, prostate cancer, and pancreatic cancer. CRC indicates colorectal cancer; ROCS, Research on Cancer Survivors.

DISCUSSION

Here we analyze cancer FH patterns in a population-based cohort of African Americans with breast cancer, prostate cancer, CRC, and lung cancer to address this gap in knowledge. The implications of cancer FH for understanding relationships between cancer types, particularly in a minority population in which FH patterns remain understudied, and for CGC referrals highlight the importance of collecting detailed cancer FHs from patients with cancer.

A large proportion of Detroit ROCS participants (47%) met NCCN guidelines for CGC referral. FH appeared to be particularly important for prostate cancer participants, with FH alone determining eligibility for 32% of participants. CGC and testing have beneficial outcomes for patients with cancer in terms of management of future cancer risk, evaluation of cancer treatment options, and risk communication with family members.^18,19 Clinical genetic testing services are more accessible than ever with the increase in the use of low-cost next-generation sequencing panels that simultaneously sequence dozens of moderate- to high-risk cancer susceptibility genes.²⁴ However, African Americans are less likely to have CGC or receive genetic testing across a spectrum of cancer types.^16,17 Although lower rates of CGC visits among African Americans could in part be due to less insurance coverage, higher medical mistrust, and concerns about cost, confidentiality, stigma, and discrimination,^25–27 lower rates of physician referrals among African Americans appear to be a primary driving factor in this disparity.^16,28 Our data clearly support the need to adhere to NCCN guidelines for CGC referrals among African American patients with cancer because nearly half of these patients may be eligible.

We also observed clustering of breast cancer and CRC, and although this is likely to partially reflect common behavioral and sociodemographic risk factors, these results are intriguing in the context of mounting evidence for increased breast cancer risk among individuals with HNPCC, who are typically at highest risk for CRC and endometrial cancer.²⁹ Women with breast cancer have recently been shown to harbor mutations in MSH2, MSH6, PMS2, and MLH1.^30–33 Lifetime breast cancer risk for MLH1 and MSH2 mutation carriers has been estimated at 18.6% to 22%,^31,32 and MSH6 and PMS2 mutation carriers have been estimated to have 2- to 3-fold increased risks of breast cancer in comparison with the general population.³³

Our observations of breast and prostate cancer FH clustering are consistent with previous reports of familial clustering of these 2 cancer types. Recently, we demonstrated in the Women’s Health Initiative prospective cohort study that an FH of prostate cancer among first-degree relatives was associated with an ~14% increased risk of breast cancer diagnosed after the age of 50 years.³⁴ We also found that African American women with an FH of both breast and prostate cancer had a 2.3-fold increase in breast cancer risk, whereas White women with a similar FH had a 1.66 times increased risk.³¹ Although BRCA1 and BRCA2 are known to increase both breast and prostate cancer risk,³⁵ the genetics underlying familial aggregation of these cancers is largely unknown.

A major strength of the Detroit ROCS cohort is the population-based identification of cases, which increases generalizability to African Americans in metropolitan Detroit and elsewhere. Our systematic ascertainment of cancer FH among the first-degree relatives and grandparents of participants also reduces the likelihood of information bias. One limitation of this study is that our FH data were self-reported only by the participants, and this could lead to misclassification of the types of cancer with which relatives were diagnosed, particularly for rarer cancer types. Mismatch repair deficiency data were also unavailable for 42% of CRC participants. We were also unable to stratify the analyses by participants’ ages at diagnosis because of insufficient sample sizes within age–cancer site strata. We also did not have FH data on specific cancers other than breast cancer, prostate cancer, lung cancer, CRC, kidney cancer, liver cancer, ovarian cancer, pancreatic cancer, and melanoma. We also had limited information on the age at diagnosis of family members because patients frequently did not recall or did not know that information (the age at diagnosis was missing for ~18% of family members). Finally, we do not currently have information on whether participants were referred for or received CGC, and in future studies, it will be important to evaluate the referral and CGC patterns among participants found to be eligible for CGC.

This study substantially adds to the current body of literature on cancer FH patterns in African Americans with breast cancer, prostate cancer, lung cancer, and CRC and highlights the importance of collecting detailed cancer FHs from individuals diagnosed with cancer. Future work to obtain more detailed rare cancer history information, to evaluate shared risk factor profiles among individuals reporting familial aggregation of different tumor types, and to perform molecular analyses of germline DNA and associated tumors will provide valuable insights into the environmental, social, and genetic etiologies of these common cancers in African Americans.