Hair product use and breast cancer incidence in the Black Women’s Health Study

Abstract

Hair relaxers and leave-in conditioners and oils, commonly used by Black/African American women, may contain estrogens or estrogen-disrupting compounds. Thus, their use may contribute to breast cancer risk. Results of the few previous studies on this topic are inconsistent. We assessed the relation of hair relaxer and leave-in conditioner use to breast cancer incidence in the Black Women’s Health Study, a nationwide prospective study of Black women. Among 50 543 women followed from 1997 to 2017, 2311 incident breast cancers occurred. Multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated using Cox proportional hazards regression for breast cancer overall and by estrogen receptor (ER) status. For heavy use (≥15 years of use for ≥7 times/year) of hair relaxers relative to never/light use (<4 years, no more than 1–2 times/year), the multivariable HR for breast cancer overall was 1.13 (95%CI: 0.96–1.33). Duration, frequency, age at first use and number of scalp burns were not associated with overall breast cancer risk. For heavy use of hair relaxers containing lye, the corresponding HR for ER+ breast cancer was 1.32 (95% CI: 0.97, 1.80); there was no association for non-lye products. There was no association of conditioner use and breast cancer. Results of this study were largely null, but there was some evidence that heavy use of lye-containing hair relaxers may be associated with increased risk of ER+ breast cancer. Consistent results from several studies are needed before it can be concluded that use of certain hair relaxers impacts breast cancer development.

We found no association of overall hair relaxer use with breast cancer risk in Black women, but there was some evidence that heavy use of lye-containing hair relaxers may be associated with an increased risk of ER+ breast cancer.

Introduction

The lifetime risk of breast cancer is similar among Black and White American women in the USA(1), but Black women are disproportionately affected by aggressive breast cancer subtypes, such as estrogen receptor (ER) negative tumors (2–4), are diagnosed at younger ages (3), and have a higher breast cancer mortality rate (1). Certain hair care products, including relaxers (straighteners) and leave-in conditioners, used more commonly by Black than White women (5–8), may contain compounds with estrogens or endocrine-disrupting chemicals (8–10). Based on estrogen’s key role in breast carcinogenesis (11–13), these products have been hypothesized to increase breast cancer risk (6,7,14,15). While the active ingredients in hair relaxers are sodium hydroxide (in lye formulations) and calcium hydroxide and guanidine carbonate (in non-lye formulations) (16), they also may contain hormonally active phthalates and parabens (14,17). Leave-in conditioners may contain placenta in addition to phthalates and parabens (5,17).

Few epidemiologic studies have assessed the relation of hair relaxer use to breast cancer risk (6,7,18–20). In the Women’s Circle of Health Study (WCHS), which included 1509 Black cases and 1290 controls recruited between 2002 and 2014, use of hair relaxers was not associated with breast cancer risk overall among Black women (6). In a case–control study in Ghana (1131 cases, 2106 controls), ever use of hair relaxers was associated with a 58% increased risk of breast cancer overall (OR = 1.58, 95% CI: 1.15–2.18), and the association was stronger for use of non-lye products than for lye products (18). In the National Institute of Environmental Health’s Sister Study, a prospective cohort of women with a family history of breast cancer, the HR for overall breast cancer among Black women for frequent use of hair relaxers in the year prior to enrollment, relative to nonuse, was 1.30 (95% CI: 0.92–1.85) (7); use of relaxers at ages 10–13 was not associated with overall risk (HR for frequent use 1.04, 95% CI: 0.77–1.40) (20).

The objective of the present analyses was to evaluate associations of hair relaxer use with breast cancer risk in Black women by ER status, with attention to patterns of use including duration, frequency, age at first use, number of scalp burns, and most commonly used type of hair relaxer (lye or non-lye). The analyses are based on data from the Black Women’s Health Study (BWHS), a large follow-up study of USA Black women. In an earlier analysis from the BWHS based on 574 incident cases occurring from 1997 through 2003, there was no association between hair relaxer use and overall breast cancer risk; ER status was not considered (19). The present analyses include 14 additional years of follow-up and 2311 cases of breast cancer.

Materials and methods

Study participants

The BWHS is a prospective cohort study established in 1995, when 59 000 self-identified Black women aged 21 through 69 years living across the USA enrolled by completing self-administered health questionnaires. The baseline questionnaire elicited information on demographic and lifestyle factors, reproductive history, and medical conditions. The cohort is followed biennially by mailed and web-based questionnaires to update exposures and ascertain incident disease, including cancer. Follow-up of the baseline cohort has been complete for >80% of potential person-years of follow-up. The study protocol was approved by the Institutional Review Board of Boston University Medical Campus. Participants indicated consent by completing and returning the questionnaires.

Follow-up for the main analyses of hair relaxer use is from 1997 through 2017. Hair relaxer use was queried in 1997 (see below). From among 52 226 participants who answered the questions on hair relaxer use, we excluded 845 women with prevalent breast cancer as of 1997, 827 women with prevalent other cancer (except non-melanoma skin cancer), and 11 women with incident breast cancer but unknown year of diagnosis, for a final analytic cohort of 50 543 women.

Case ascertainment

Incident cases of breast cancer were ascertained through self-report on biennial follow-up questionnaires (89% of cases) or through death records or linkage to 24 cancer registries in states covering 95% of participants (11% of cases). Breast cancer diagnoses were confirmed by review of medical records, pathology reports, and cancer registry records. Data on tumor characteristics were abstracted from these records.

Among the 50 543 women in the analytic cohort followed from 1997 to 2017, 2311 women developed incident breast cancer over follow-up, including 1843 invasive breast cancers and 468 ductal carcinoma in situ (DCIS). Among the 2311 invasive and DCIS cases, there were 1420 ER+, 601 ER–, and 286 triple negative (ER–, progesterone receptor (PR) –, HER2–) tumors.

Assessment of hair product use

The 1997 questionnaire ascertained hair relaxer use with six questions (response categories): ‘Have you ever used a chemical hair straightener?’ (yes, no); ‘At what age did you first use chemical hair straighteners?’ (less than 10, 10–19, 20–29, 30 or more years old); ‘How often do you (or did you) use chemical hair straighteners?’ (about 1, 2, 3–4, 5–6, 7 or more times/year); ‘In total, how many years have you used hair straighteners?’ (less than 1, 1–4, 5–9, 10 or more years); ‘How many times have you experienced burns (a break in the skin, not just tingling) during the application of chemical straighteners?’ (never, 1–2, 3–4, 5–9, 10 or more times); ‘Which of the following chemical hair straighteners have you used most often?’ (lye, no-lye, don’t know).

On the 2009 questionnaire, we asked about ever use, and years of use, of ‘Infusium 23 leave-in conditioner,’ ‘Hask Placenta hot oil treatment,’ ‘Organic Root Stimulator—Olive Oil,’ and ‘other’. We also asked which product was most often used during the teenage years: Infusium 23, Hask Placenta hot oil treatment, or other.

Covariate assessment

The 1995 baseline questionnaire collected information on established and putative risk factors for breast cancer including adult height, weight, weight at age 18, age at menarche, oral contraceptive use, parity, age at first birth, menopausal status, postmenopausal hormone use, alcohol consumption, physical activity, educational attainment, breast cancer in first-degree relatives, and history of mammograms. Except for height, age at menarche and weight at age 18, all data were updated on subsequent questionnaires.

Statistical analysis

In the hair relaxer analyses, women contributed person-years from the beginning of follow-up in 1997 until diagnosis of breast cancer, other cancer, death, or end of follow-up in March 2017, whichever occurred first (943 392 total person-years). Mean follow-up time was 17.8 years (standard deviation, 4.8). We used Cox proportional hazards regression models, stratified by age in one-year intervals and questionnaire cycle such that age was the underlying time scale, to estimate HRs and 95% CIs for use of hair relaxers in relation to risk of overall, ER+, ER–, and triple negative breast cancer. The analyses combined invasive and DCIS cases for consistency with prior reports (6,7,19,20).

Because few women reported never using hair relaxers, we used a reference group for all comparisons that included light users of any type of hair relaxers (i.e., used for <4 years at a frequency of no more than 1–2 times/year) in addition to never users. We estimated HRs for ever use of hair relaxers, heavy use (15 or more years with a frequency of ≥7 times/year), duration of use, frequency of use, age at first use, number of burns, and whether lye or non-lye relaxer was most commonly used. In addition to age and questionnaire cycle, multivariable models adjusted for geographic region of residence (South, Midwest, Northeast, and West), oral contraceptive use (never, ever), age at menarche (<11, 11, 12–13, 14+ years), body mass index at age 18 (<20, 20–24, 25+ kg/m²), current body mass index (<25, 25–29, 30–34, ≥35 kg/m²), and number of mammograms received up to four years prior to breast cancer diagnosis (0, 1–4, ≥5). These covariates were chosen because of observed associations with hair relaxer use (i.e., geographic region, oral contraceptive use, and current BMI) and/or because they are strong breast cancer risk factors (i.e., oral contraceptive use, age at menarche, BMI at age 18); we included mammography in the model to control for possible detection bias. Where applicable, covariates were treated as time-varying in the models. Results from models adjusted only for age and questionnaire cycle and from the multivariable-adjusted models were similar, and we present multivariable results in the tables. Further adjustment for other breast cancer risk factors (parity, age at first birth, menopausal status, alcohol consumption, physical activity, and use of postmenopausal hormones) did not change the HRs.

We tested for linear trend across ordinal variables for duration of use, frequency of use, and number of burns. To evaluate possible effect modification, we estimated HRs within strata of age (<45, ≥45 years), history of breast cancer in a first degree relative (yes/no), BMI (<30, ≥30 kg/m²), and menopausal status. We tested for the significance of any differences in the HRs for hair relaxer use in strata of each of the stratification variables using the -2 log likelihood test.

In further analyses, we evaluated associations of use of hair oils or leave-in conditioners with risk of breast cancer overall and by ER status. For these analyses, women contributed person-years from 2009, when participants were asked about use of these products, until diagnosis of breast cancer, other cancer, death, or end of follow-up in March 2017. We estimated HRs for ever use of any hair conditioning product and for ever use and duration in years of use for each of the three specific products queried (Duration was not queried for use of ‘other’ products.) The reference group was never use of any of these products. For analyses of products used ‘most often in your teenage years’, follow-up began in 1997 because teenage use would have predated baseline for all women. The reference group was never use of any hair conditioning products, as defined above. Women who gave contradictory answers to the hair conditioner questions were excluded from the analyses (n = 4130).

Results

Hair relaxers

Five percent of participants never used hair relaxers, 5% were light users (1–4 years, ≤2 times/year), 70% were moderate users (1–4 years, >3 times/year; 4–14 years, any frequency; and ≥15 years, <7 times/year), and 20% were heavy users (≥15 years, ≥7 times/year). Among the moderate and heavy users, 37% used them for ≥20 years, 33% used them ≥7 times/years, 34% reported at least 5 burns, and 64% first started using relaxers before age 20. Most participants (71%) most commonly used non-lye relaxers, 16 % most commonly used lye relaxers and 13% did not know what type they used.

Table 1 shows the age-adjusted distribution of selected participant characteristics in 1997 among never/light users of hair relaxers, moderate users, and heavy users. Compared to never/light users, heavy users were younger, less likely to have BMI <25 kg/m², less likely to engage in physical activity, and more likely to live in the South or Midwest.

Table 1.

Age-Standardized Characteristics of 50,543 BWHS participants in 1997 according to hair relaxer use^a,b

	Never or light use (n = 5183)	Moderate use (n = 35,194)	Heavy use(n = 10,166)
Age (mean ± SD)	44.6 ± 12.0	40.5 ± 10.7	40.0 ± 9.2
Column %s
BMI at age 18, kg2/m
<20	42	40	40
20–24	43	44	45
≥25	13	14	14
Body Mass Index, kg/m2
<25	41	34	33
25–29	31	32	34
30–34	15	18	18
≥35	12	15	15
Parity
Nulliparous	32	32	33
1 child	22	23	23
2 children	23	24	25
≥3 children	22	21	19
Age at first birth (among parous women)
<20 years	31	32	32
20–24 years	36	34	35
≥25 years	31	32	32
Alcohol consumption, drinks/week
<1	74	72	72
1–6	21	23	23
≥7	3	4	4
Physical activity
< 1 h/week	58	60	63
1–4 hs/week	29	28	27
≥5 h/week	10	9	8
Age at menarche
≤11 years	26	29	29
12–13 years	53	52	52
≥14 years	21	19	18
Family history of breast cancer	10	9	9
Menopausal status
Premenopausal	74	74	73
Postmenopausal	19	19	20
Educational attainment (1995)
≤12 years	22	18	18
13–15 years	34	36	35
16 years	22	24	26
≥17 years	22	22	22
Region of residence (U.S.)
Northeast	28	28	22
South	24	31	38
Midwest	20	23	25
West	28	18	15
Ever use of postmenopausal hormones	4	4	4
Ever use oral contraceptives	80	85	88
Had mammogram between 1995 and 1997	51	53	53

^aAge standardized in 5 year intervals; all characteristics from 1997 unless noted otherwise; percentages may not add to 100 due to missing values.

^bNo or light use is 1–4 years, ≤2 times/year; heavy use is use for ≥15 years, ≥7 times/year; remainder are moderate users (1–4 years, >3 times/year; 4–14 years, any frequency; and ≥15 years, <7 times/year).

Table 2 shows multivariable HRs for the associations of hair relaxer use with breast cancer risk overall and by ER status. There were no significant trends in HRs with increases in years of use, frequency of use, age at first use, or number of burns (all P for trend >0.05). For breast cancer overall, compared with never/light use the multivariable-adjusted HR for heavy use was 1.13 (95% CI: 0.96–1.33) for heavy use. HRs for all exposure metrics were similar for invasive and DCIS cases considered separately. For example, the HRs for heavy use of hair relaxers relative to never/light use were 1.15 (95% CI: 0.96–1.38) and 1.03 (95% CI: 0.72–1.48) for invasive breast cancer and DCIS, respectively.

Table 2.

Associations of hair relaxer use with breast cancer, overall and by ER status, 1997–2017

	Person-years	All Breast Cancer (n=2311)				ER positive (n=1420)				ER negative (n=601)
		cases	HRa	95% CI		cases	HRa	95% CI		cases	HRa	95% CI
Hair relaxer use
Never/light useb	94,599	252	1.00	reference		152	1.00	reference		59	1.00	reference
Ever use	848,793	2059	1.05	0.92	1.20	1268	1.08	0.91	1.28	542	1.11	0.84	1.46
Heavy usec	191,248	491	1.13	0.96	1.33	298	1.15	0.93	1.42	138	1.21	0.88	1.67
Duration of use, years
1–4	37,933	117	1.21	0.97	1.51	74	1.26	0.95	1.66	29	1.26	0.81	1.97
5–9	96,574	192	0.95	0.78	1.14	111	0.90	0.70	1.15	53	1.08	0.74	1.56
10–14	193,471	361	1.04	0.88	1.22	225	1.06	0.86	1.31	93	1.06	0.76	1.48
15–19	211,224	477	1.14	0.97	1.33	292	1.16	0.94	1.42	133	1.24	0.91	1.70
≥20	309,591	912	1.02	0.89	1.18	566	1.06	0.89	1.28	234	1.06	0.79	1.41
Frequency of use/yeard
1–2	63,514	160	0.96	0.79	1.17	93	0.94	0.73	1.21	46	1.11	0.76	1.62
3–4	245,745	645	1.05	0.91	1.22	390	1.06	0.88	1.27	175	1.13	0.85	1.50
5–6	254,345	582	0.99	0.86	1.15	369	1.06	0.88	1.28	140	0.92	0.68	1.24
≥7	278,863	655	1.12	0.97	1.30	407	1.17	0.97	1.42	175	1.13	0.85	1.52
Age at first use
<10	48,035	73	1.02	0.77	1.35	42	0.96	0.67	1.38	21	1.05	0.62	1.79
10–19	501,839	1,029	1.11	0.9	1.31	632	1.14	0.93	1.42	277	1.06	0.77	1.45
20–29	240,349	740	1.08	0.9	1.28	460	1.14	0.92	1.40	184	1.02	0.74	1.40
≥30	57,339	207	0.95	0.78	1.16	128	0.97	0.75	1.25	59	1.12	0.76	1.65
Number of burns
Never burned	135,310	352	1.00	0.84	1.19	205	0.97	0.77	1.22	95	1.02	0.72	1.44
1–2	234,523	583	1.06	0.90	1.26	361	1.11	0.90	1.38	154	1.03	0.74	1.53
3–4	184,092	449	1.11	0.93	1.32	279	1.16	0.93	1.45	119	1.07	0.76	1.50
5–9	119,302	266	1.07	0.88	1.29	170	1.16	0.91	1.47	64	0.93	0.63	1.36
>10	173,767	400	1.12	0.94	1.34	248	1.16	0.93	1.46	108	1.08	0.76	1.54

^aAdjusted for age, questionnaire cycle, region (south, midwest, northeast, west), oral contraceptive use (never, ever, UK),

body mass index (<25, 25–29, 30–34, 35+), age menarche (<11, 11, 12–13, 14+), body mass index at age 18 (<20, 20–24, 25+, UK), and number of mammograms received up to 4 years prior to breast cancer diagnoses (0, 1–4, 5+).

^bReference group includes never users and those who used hair relaxers for <4 years at a frequency no more than 2 times/year; it is the reference group for all exposures in the table.

^cA subset of ever use: used >=15 years, 7+ times/year.

^dP for trend over frequency of use for ER+ cancer was 0.03. All other P for trend were >0.05.

In general, results were similar for ER+ and ER– breast cancer (Table 2): the HRs for heavy use of hair relaxers were 1.15 (95% CI: 0.93–1.42) and 1.21 (95% CI: 0.88–1.67), respectively. As in the overall analyses, there were no apparent trends in patterns of use (all P for trend >0.05) with one exception: the P for trend over frequency of use for ER+ cancer was 0.03. Results for triple negative breast cancer were similar to those for ER– breast cancer (Supplementary Table 1, available at Carcinogenesis Online).

Table 3 shows results separately for women who most commonly used lye relaxers and for women who most commonly used non-lye relaxers. The HRs for overall breast cancer, ER+ breast cancer, and ER– breast cancer associated with heavy use of lye products were 1.22 (95% CI: 0.95–1.56), 1.32 (95% CI: 0.97–1.80) and 1.27 (95% CI: 0.79–2.02), respectively. HRs did not increase as the duration of use increased, but did increase as frequency of use increased, with the highest HR seen for ER+ cancer (HR for ≥7 times/year: 1.37 (95% CI:1.04–1.82)); the P for trend was 0.06. HRs were also highest among women who reported ≥10 burns, though no HRs were statistically significant. Among women who most commonly used non-lye hair relaxers, there were no material increases in HRs by duration, frequency of use, or number of burns (all P for trend >0.05).

Table 3.

Association of breast cancer, overall and by ER status,among users of lye and non-lye hair relaxer

	All Breast Cancer				ER+				ER-
	cases	HRa	95% CI		cases	HRa	95% CI		cases	HRa	95% CI
			Most often used lye hair relaxer
Never/light useb	252	1.00	reference		152	1.00	reference		59	1.00	reference
Ever use	334	1.10	0.94	1.31	214	1.18	0.96	1.16	83	1.09	0.78	1.53
Heavy usec	105	1.22	0.95	1.56	67	1.32	0.97	1.8	30	1.27	0.79	2.02
Duration of use in years
1–9	44	1.22	0.88	1.69	28	1.29	0.85	1.94	9	0.98	0.48	1.98
10–19	150	1.25	1.00	1.56	101	1.38	1.05	1.83	35	1.12	0.72	1.75
≥20	140	1.00	0.80	1.23	85	1.03	0.78	1.37	39	0.96	0.64	1.47
Frequency of use per year
1–4	86	0.98	0.76	1.26	52	0.99	0.72	1.36	24	1.04	0.65	1.69
5–6	107	1.07	0.84	1.35	70	1.17	0.87	1.57	22	0.79	0.48	1.3
≥7	138	1.26	1.01	1.58	90	1.37	1.04	1.82	36	1.19	0.77	1.83
Number of burns
Never burned	34	0.96	0.67	1.38	19	0.9	0.56	1.46	10	1.06	0.54	2.08
1–9	199	1.07	0.88	1.30	129	1.16	0.9	1.48	49	0.98	0.66	1.45
>10	102	1.23	0.97	1.57	66	1.32	0.97	1.79	25	1.12	0.69	1.82
	Most often used non-lye hair relaxer
	casesa	HRa	95% CI		casesa	HRa	95% CI		casesa	HRa	95% CI
Never/light use	252	1.00	Reference		152	1.00	Reference		59	1.00	Reference
Ever use	1412	1.03	0.90	1.18	866	1.05	0.88	1.26	388	1.14	0.86	1.50
Heavy usec	328	1.13	0.95	1.35	200	1.18	0.94	1.49	90	1.13	0.8	1.59
Duration of use in years
1–9	198	1.01	0.84	1.22	121	1.02	0.8	1.3	55	1.13	0.78	1.64
10–19	577	1.05	0.90	1.23	347	1.06	0.87	1.29	164	1.17	0.86	1.59
≥20	637	1.04	0.89	1.20	398	1.1	0.91	1.33	169	1.09	0.8	1.47
Frequency of use per year
1–4	581	1.05	0.90	1.21	349	1.05	0.87	1.27	167	1.22	0.91	1.64
5–6	384	0.95	0.81	1.11	237	0.98	0.8	1.21	104	1.01	0.73	1.39
≥7	436	1.11	0.95	1.30	273	1.17	0.96	1.43	115	1.13	0.82	1.55
Number of burns
Never burned	248	0.95	0.79	1.13	147	0.94	0.75	1.18	69	1.08	0.77	1.54
1–9	919	1.05	0.91	1.22	567	1.1	0.91	1.32	253	1.15	0.86	1.53
>10	242	1.04	0.86	1.24	150	1.09	0.86	1.37	66	1.09	0.76	1.56

^aAdjusted for age, questionnaire cycle, region (south, midwest, northeast, west), oral contraceptive use (never, ever, UK), current body mass index (<25, 25–29, 30–34, 35+), age menarche (<11, 11, 12–13, 14+), body mass index at age 18 (<20, 20–24, 25+, UK), and number of mammograms received up to four years prior breast cancer diagnosis (0, 1–4, 5+).

^bNever used, or used for 1–4 years at a frequency of no more than 2X/yr; Reference for all comparisons.

^cUsed >=15 years, 7+ times/yr.

We assessed associations by family history of breast cancer (Supplementary Table 2, available at Carcinogenesis Online). Among women with no family history, the HR for heavy use of any hair type of hair relaxer was 1.21 (95% CI:1.01–1.45) and the HR for heavy use of lye hair relaxers was 1.45 (95% CI:1.11–1.88). In contrast, among women with a positive family history, the corresponding HRs were 0.41 (95% CI:0.56–1.26) and 0.41 (95% CI:0.16–1.02); the latter result was based on only 6 exposed cases (P for interaction<0.01). There were no material differences in HRs by age or menopausal status (Supplementary Table 2, available at Carcinogenesis Online).

Finally, we estimated HRs for heavy use of any hair relaxer and lye-containing relaxers using the smaller reference group of women who never used hair relaxers. Results with the restricted reference group were similar to our main results: the HR for heavy use of any hair relaxer was 1.05 (95% CI: 0.85–1.30) and HR for heavy use of lye-containing relaxers was 1.10 (95% CI: 0.82–1.47).

Hair conditioners and oils

Twenty-seven percent of women ever used Infusium 23, 28% of whom used it >5 years; 9% ever used Hask oil, 24% of whom used it >5 years; and 25% used some other type of oil, 15% of whom used it >5 years. Women who never used hair conditioners or oils were older than those who had ever used them (mean age 42.6 years compared to 39.6 years). The distributions of other covariates were similar for never/light and ever users (data not shown).

There was no association of ever use of leave-in hair conditioners, lotions, or oils with breast cancer overall (HR = 1.00, 95% CI:0.86–1.18 for any use versus no use) (Supplementary Table 3, available at Carcinogenesis Online). Compared to women who did not report use of any of the conditioners or oils, the HR for all breast cancer associated with ≥5 years of use of Infusium 23 was 1.13 (95% CI:0.90–1.43), for ≥3 years of use of Hask Placenta Hot Oil was 0.72 (95% CI:0.50–108), and for ≥5 years of use of Organic Root Stimulator was 0.98 (95% CI:0.70–1.36). All P for trend >0.05 with the exception of duration of hask oil use for all breast cancers (P trend = 0.03). Results were similarly null for ER+, ER–, and triple negative breast cancer (Supplementary Table 3, available at Carcinogenesis Online). There was also no apparent association with use of these products during the teenage years (Supplementary Table 4, available at Carcinogenesis Online).

Discussion

In this large cohort of Black women, we found little evidence to support an association of overall hair relaxer use with risk of breast cancer. There was, however, some evidence of an association of heavy use of lye formulations with increased risk of ER+ breast cancer: HRs for these formulations were consistently elevated for heavy users, frequent users, and women who reported ≥10 burns, though most increases were not statistically significant. Use of non-lye hair relaxers was not associated with increased breast cancer risk.

The use of hair relaxers is common among Black women (5,8). These products are hypothesized to impact breast cancer risk via exposure to hormonally active compounds (8–10) through dermal absorption or scalp burns. Both lye and non-lye relaxers are likely to contain endocrine disrupting chemicals, including phthalates used for fragrance and parabens used as preservatives (14,17). Hair relaxers may contain formaldehyde, a carcinogen (21). The specific formulations are proprietary, and every study carried out to date has lacked important information not only on their constituents but on which specific products were used. Levels of urinary metabolites of parabens and phthalates have been shown to be higher in Black than White women (22–24). Parabens and phthalates have also been detected in human breast tumors (25,26). Accumulating evidence from experimental (27–30) and animal (31,32) studies supports the carcinogenic potential of these chemicals, and increased risk of breast cancer has been associated with exposure to known endocrine disruptors (33) and higher levels of urinary parabens (34) and phthalates (35) in some but not all (36,37) population-based studies. Use of relaxers and hair oils has been associated with early menarche (38), a risk factor for breast cancer (39). We found stronger associations among women who most commonly used lye versus non-lye products. While non-lye relaxers are advertised to cause fewer scalp burns than lye relaxers (40), burns occurred among both non-lye and lye users in our study.

Two previous studies considered lye and non-lye hair relaxers separately. Among Black women in the WCHS, the OR for lye formulations was 1.19 (95% CI: 0.92–1.54) and for non-lye formulations the OR was 0.84 (95% CI: 0.66–1.08).(6) In contrast to our findings, in the WCHS study lye formulations were associated with ER– breast cancer (OR = 1.54, 95% CI: 0.93–2.53) rather than ER+ breast cancer (OR = 0.94, 95% CI: 0.69–1.28) (6). In the Ghana case-control study, a higher risk of breast cancer was found for use of non-lye (OR = 1.88, 95% CI: 1.34–2.64) than for lye hair relaxers (OR = 1.38, 95% CI: 0.99–1.93) and there was no trend for number of burns; analyses by ER status were not performed (18). The lack of consistency in results among the WCHS, Ghana, and BWHS studies could be due in part to misclassification of product type: the majority of women who used hair relaxers in the BWHS reported ‘most often’ using non-lye products but use of lye and non-lye products were likely not mutually exclusive and an appreciable proportion of participants did not know which type they most commonly used. In addition, there could be differences in types of products used among the three studies due to differences in geography and time periods over which hair relaxer use was reported. In the most recent study, the Sister Study, type of hair relaxer was not queried (7).

In the Sister Study, frequent use of hair relaxers was associated with nonsignificant increases in breast cancer risk in both Black (HR = 1.30, 95% CI: 0.92–1.85) and White (1.26, 95% CI: 0.88–1.80) women. All participants had a first degree relative with breast cancer (7). In the BWHS, evidence of a positive association was most apparent among women who had not reported having a first degree relative with breast cancer. In additional analyses in the Sister Study, frequent hair straightener use during adolescence was associated increased risk of premenopausal breast cancer (HR = 2.11, 95% CI: 1.26–3.55) but not postmenopausal breast cancer (HR = 0.99, 95% CI: 0.76–1.30) (20). In the present analyses, there were no differences in the HRs by menopausal status or by age.

We also assessed use of deep hair conditioners and oils in relation to breast cancer incidence. The available data were less detailed than those for hair relaxer use and were collected later in follow-up, providing less statistical power to detect associations. Consistent with the WCHS report by Llanos et al. (6). we found no evidence that use of these products increases breast cancer risk. Again, interpretation of this result is limited given potential misclassification of exposure and non-exclusive use of the specific products queried.

The strengths of the present study include its prospective design, detailed assessment of patterns of hair relaxer use, and ability to control for potential confounding variables. In addition, this is the largest study of hair relaxer use in relation to breast cancer in women of African ancestry to date. Exposure misclassification could have limited our ability to detect associations. Hair relaxer use was not updated subsequent to 1997, so changes in use patterns over time were not captured. In addition, we lacked a truly unexposed reference group. Ninety-five percent of BWHS participants reported having used hair relaxers; in order to achieve stability in effect estimates, we included women who reported infrequent or light use of these products in the reference group, which may have biased HRs to the null. Of note, however, the point estimates of the HRs were similar when we restricted the reference group to never users. We did not have information on specific brands of hair relaxers or formulations; to the extent that formulations have changed over time, results from our study may only be generalizable to use of products on the market before 1997.

In conclusion, the results of our study were largely null, although there was some evidence for the possibility of increased risk of ER+ breast cancer among women who were heavy users of lye-containing relaxers. While there is biologic plausibility that exposure to some components contained in hair relaxers might increase breast cancer risk, the evidence from epidemiologic studies to date is inconsistent. Given the high prevalence of use of these products among Black women, further research is warranted to better establish associations according to breast cancer subtype as well as according to specific products.

Funding

This work was supported by the National Institutes of Health (U01 CA164974, R01 CA58420).

Glossary

Abbreviations

BWHS

Black Women’s Health Study

CIs

confidence intervals

DCIS

ductal carcinoma in situ

ER

estrogen receptor

HRs

hazard ratios

WCHS

Women’s Circle of Health Study

Conflict of interest

None declared.

Data Availability

The data underlying this article cannot be shared publicly for the privacy of individuals that participated in the study. The data will be shared on reasonable request to the corresponding author.