Hysterectomy, oophorectomy and risk of Non-Hodgkin’s lymphoma

Abstract

Hysterectomy is associated with an increased risk for adverse health outcomes. However, its connection to the risk of non-Hodgkin’s lymphoma (NHL) remains unclear. The aims of this study were to investigate the associations between hysterectomy, oophorectomy and risk of NHL and its major subtypes (e.g. diffuse large B-cell lymphoma (DLBCL)), and whether these associations were modified by exogenous hormone use. Postmenopausal women (n=141,621) aged 50–79 years at enrollment (1993–1998) from the Women’s Health Initiative were followed for an average of 17.2 years. Hysterectomy and oophorectomy were self-reported at baseline. Incident NHL cases were confirmed by central review of medical records and pathology reports. During the follow-up period, a total of 1,719 women were diagnosed with NHL. Hysterectomy, regardless of oophorectomy status, was associated with an increased risk of NHL (hazard ratio [HR] = 1.23, 95% CI: 1.05 – 1.44). Oophorectomy was not independently associated with NHL risk after adjusting for hysterectomy. When stratified by hormone use, the association between hysterectomy and NHL risk was confined to women who had never used hormone therapy (HR=1.35, 95% CI: 1.06–1.71), especially for DLBCL subtype (p for interaction=0.01), and to those who had undergone hysterectomy before the age of 55. Our large prospective study showed that hysterectomy was a risk factor of NHL. Findings varied by hormone use. Future studies incorporating detailed information on the types and indications of hysterectomy may deepen our understanding of the mechanisms underlying DLBCL development and its potential interactions with hormone use.

Graphical Abstract

Background

Non-Hodgkin’s lymphoma (NHL) is the seventh most common malignancy in the United States, accounting for about 4% of all cancers.¹ The established risk factors for NHL as a single entity include being male, of White race, having family history, radiation exposure, exposure to certain chemicals, and some autoimmune diseases such as rheumatoid arthritis or systemic lupus erythematosus.^{2, 3} High BMI in early adulthood is also associated with a higher risk of NHL.^3–5

NHL consists of many histologically and biologically distinct subtypes. Diffuse large B-cell lymphoma (DLBCL) is the most common NHL subtype, constituting 25–30% of adult NHL cases in the US, followed by follicular lymphoma (FL) accounting for approximately 15–20% of adult NHL. Chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL) is another common subtype of NHL in Western countries but rare in Asian countries. The incidence of NHL subtypes varies by age, sex, race/ethnicity, and geographic region.⁶ Few prior studies have evaluated NHL-subtype-specific risk factors.^6–8

NHL incidence is significantly higher in males than in females worldwide; however, the mechanisms underlying the sex difference are not yet understood. One possible mechanism may be related to the effects of estrogens on lymphoma cell proliferation or on the host immune response.^9–11 However, epidemiologic evidence on hormonal factors and risk of NHL is inconsistent. While case-control studies have tended to report lower risks of NHL among menopausal hormone users,¹² cohort studies have yielded more mixed results ranging from significantly increased risk^{13, 14} to significantly lower risk for the DLBCL subtype to null associations.^15–17 The results from randomized clinical trials from the Women’s Health Initiative do not support a role of estrogen alone or combined with progestin in the development of NHL among postmenopausal women.¹⁸

Of note, two studies have suggested an increased risk of B-cell NHL associated with hysterectomy plus bilateral salpingo-oophorectomy (BSO). The California Teachers Study reported hysterectomy plus BSO was associated with significantly higher risk of B-cell NHL among women who had never used menopausal hormone therapy (RR=3.15 95% CI=1.62–6.13).¹⁵ Similarly, the European Prospective Investigation Into Cancer and Nutrition study reported that women who had undergone surgical menopause due to bilateral ovariectomy had a 51% significantly higher risk of B-cell NHL than women with natural menopause.¹⁹ However, both studies were based on small numbers of cases. Hysterectomy is a commonly performed surgical procedure in women.²⁰ BSO is often performed at the time of hysterectomy. Since premenopausal BSO suddenly reduces both estrogen and androgen productions,²¹ unopposed estrogen, which refers to estrogen therapy without the addition of progesterone or progestin, is typically recommended for women who have had a hysterectomy, especially if the ovaries are also removed. Thus, it is important to examine the association between hysterectomy and risk of NHL separated by oophorectomy status and by exogenous hormone use to shed light on the associations of endogenous or exogenous hormones with the risk of NHL.

In the present study, we used the Women’s Health Initiative (WHI), a large prospective study in the US, with detailed information on potential confounders and centrally-adjudicated NHL cases and subtypes, to investigate the association between hysterectomy, oophorectomy and risk of NHL and its major subtypes, including DLBCL, FL and CLL/SLL. We further examined whether the association between hysterectomy, BSO and risk of NHL was modified by unopposed estrogen use. We hypothesized that undergoing a hysterectomy would be associated with higher risk of NHL, particularly when combined with BSO, and that utilization of unopposed estrogen after hysterotomy would modify this association.

Methods

Women’s Health Initiative

The WHI is a large prospective study among US postmenopausal women.²² Details of the study’s design and participant recruitment are described elsewhere²³. Briefly, 161,808 women ages 50 to 79 years were recruited from 40 clinical centers between 1993 and 1998. The study included four overlapping clinical trials (CT) (including estrogen alone trial, estrogen plus progestin trial, dietary modification trial and calcium and vitamin D trial) and an observational study (OS). Participants in the OS were followed annually, while those in the clinical trials were followed-up every 6 months through 2005 and annually thereafter. During the original study concluded in 2005, 4% were lost to follow-up or stopped to follow-up, and 5% were deceased. After the completion of the initial protocol in 2005, further outcome assessments required re-consent, which was obtained from a majority (more than 77%) of surviving and active participants (115,407 out of 150,076) (extension 1: 2005–2010). In 2010, 87% (93,567 out of 107,706 women) again consented to extended follow-up (extension 2: 2010–2020).²⁴ A WHI estrogen plus progestin trial study showed that the baseline characteristics were similar by re-consent status with slightly higher participation by White women compared to other ethnic groups.²⁵

For this analysis, all 161,808 participants in the WHI cohort were considered. However, certain exclusions were made, including 14,849 women with a history of cancer (excluding non-melanoma skin cancer) at baseline, 630 with no follow-up information, 1,538 women with missing information on hysterectomy, age of hysterectomy, or oophorectomy collected at enrollment, and 3,170 women with missing information on important factors such as hormone use, BMI, and diet quality. As a result, 141,621 women were included in the subsequent analysis (Figure 1). All of these participants were followed up to the date of NHL diagnosis, death, loss-to-follow-up, or the end of follow-up (February 19,2022), whichever came first.

Figure 1.

Flow diagram of participants included in the final analyses.

Measurements

Outcome ascertainment

The focus of our study was the occurrence of NHL as the primary outcome. Newly diagnosed NHL cases during follow-up were initially identified by self-reports through mail or telephone questionnaires. These cases were then further confirmed by centralized trained cancer adjudicators who meticulously reviewed medical records and pathology reports obtained from hospitals and/or laboratories.²⁶ Classification and histology of tumors in the WHI were based on the SEER guidelines; the case morphology classification codes used ICD-O-3 coding.²⁶ For this study, we specifically focused on three major subtypes of NHL: diffuse large B-cell lymphoma (DLBCL), follicular lymphoma (FL), and chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL) .²⁷

Exposures

Surgical histories of hysterectomy and oophorectomy and age at surgery were collected through these self-reported questionnaires.

Hysterectomy and oophorectomy status

Hysterectomy status at baseline was determined by asking participants the question, “Have you ever had a hysterectomy? (This is a surgery to remove your uterus or womb.)” Oophorectomy status at baseline was assessed based on whether participants had undergone surgery to remove one or both ovaries. The responses were categorized as follows: no, yes (one ovary removed), yes (both ovaries removed), yes (part of an ovary removed), yes (unknown number removed), and don’t know. Additionally, the age(s) at which hysterectomy and oophorectomy occurred were collected during the baseline assessment. It is worth noting that the accuracy of self-reported hysterectomy and oophorectomy has been previously described in a similar population, with a sensitivity of 91% and positive predictive value of 97% for hysterectomy status, and a sensitivity of 73% and positive predictive value of 100% for oophorectomy status.²⁸

During the follow-up period, women were asked about their hysterectomy status by inquiring, “Since your last medical update, which of the following exams, tests, or procedures have you had?” Information on updated hysterectomy status was obtained based on the response option indicating “removal of the uterus or womb (hysterectomy).” However, information on oophorectomy status during follow-up was not collected.

Exogenous hormone use

At baseline, information on the lifetime use of menopausal hormones was gathered through structured questionnaires and visual aids comprising colored photographs depicting different hormone preparations. The participants’ history of using exogenous hormones was categorized into four groups: none, estrogen alone, estrogen and progestin, and mixed.

Covariates

We developed a directed acyclic graph (DAG) (Figure 2) to assist in the identification of appropriate covariates, and subsequently considered the following potential risk factors measured at baseline, including demographics (age in years, race, ethnicity, education), family history of any cancer, lifestyle factors (diet quality, body mass index (BMI), smoking, physical activity and alcohol consumption), exogenous hormone use, several medical conditions linked to NHL (rheumatoid arthritis or lupus), and participation in different WHI study subcohorts (observational study or clinical trials and different treatment assignments for all 4 clinical trials). Race and ethnicity were self-reported. We categorized race as: White, Black, Asian, and other; and ethnicity as not Hispanic, Hispanic or missing. To assess diet quality, we utilized the Food Frequency Questionnaire and calculated the Healthy Eating Index (HEI-2015) score based on the dietary guidelines provided in 2015 for Americans.²⁹ Physical activity was measured in metabolic equivalent task (MET)-hours per week. We gathered information on smoking habits, including current, former, or never smokers, as well as details on smoking intensity and duration. Pack-years were estimated for former and current smokers.

Figure 2.

A Directed Acyclic Graph (DAG) to determine the appropriate covariates for adjustment

Data analysis

Baseline characteristics were summarized as percentages for categorical variables and means (standard deviation) for continuous variables. To compare women across the combinations of hysterectomy and oophorectomy status, the Chi-square test was used for categorical variables, while ANOVA test was employed for continuous variables (Table 1).

Table 1.

Baseline characteristics of study participants (N=141,621) by hysterectomy status in the Women’s Health Initiative, United States, 1993–1998

	Hysterectomy
Variable	Overall N=141,621	No hysterectomy and no oophorectomy N=81,243	No hysterectomy but with oophorectomy N=3,933	Hysterectomy without oophorectomy N=20,853	Hysterectomy with oophorectomy N=35,592
Age at baseline	63.0 ± 7.2	62.9 ± 7.2	63.4 ± 7.4	62.9 ± 7.2	63.4 ± 7.2
Race
Asian	3,678 (2.6%)	2,373 (2.9%)	92 (2.3%)	357 (1.7%)	856 (2.4%)
Black	12,331 (8.7%)	5,347 (6.6%)	305 (7.8%)	2,269 (10.9%)	4,410 (12.4%)
White	120,722 (85.2%)	70,874 (87.2%)	3,402 (86.5%)	17,421 (83.5%)	29,025 (81.5%)
Other	4,890 (3.5%)	2,649 (3.3%)	134 (3.4%)	806 (3.9%)	1,301 (3.7%)
Hispanic ethnicity
No	134,142 (94.7%)	77,062 (94.9%)	3,746 (95.2%)	19,603 (94.0%)	33,731 (94.8%)
Yes	6,333 (4.5%)	3,500 (4.3%)	154 (3.9%)	1,117 (5.4%)	1,562 (4.4%)
Missing	1,146 (0.8%)	681 (0.8%)	33 (0.8%)	133 (0.6%)	299 (0.8%)
Education
High school or less	31,559 (22.4%)	16,497 (20.4%)	828 (21.2%)	5,238 (25.3%)	8,996 (25.5%)
Some college/technical training	53,226 (37.9%)	28,669 (35.5%)	1,431 (36.7%)	8,673 (41.9%)	14,453 (40.9%)
College or some post-college	3,1469 (22.4%)	19,712 (24.4%)	911 (23.3%)	4,006 (19.4%)	6,840 (19.4%)
Master or higher	24,330 (17.3%)	15,805 (19.6%)	733 (18.8%)	2,775 (13.4%)	5,017 (14.2%)
Pack-years of smoking
Never smoker	71,888 (52.5%)	40,885 (52.0%)	1,762 (46.4%)	10,947 (54.2%)	18,294 (53.0%)
<5	19,878 (14.5%)	11,357 (14.5%)	540 (14.2%)	2,965 (14.7%)	5,016 (14.5%)
5–<20	19,755 (14.4%)	11,453 (14.6%)	610 (16.1%)	2,767 (13.7%)	4,925 (14.3%)
≥20	25,518 (18.6%)	14,870 (18.9%)	884 (23.3%)	3,510 (17.4%)	6,254 (18.1%)
Body-mass Index (BMI), kg/m2	28.0 ± 5.9	27.6 ± 5.9	27.8 ± 5.8	28.4 ± 5.8	28.6 ± 6.0
Physical activity (MET-hours/wk) 1	12.5 ± 13.7	13.1 ± 14.0	12.9 ± 14.0	11.7 ± 13.4	11.5 ± 13.2
Diet quality (HEI 2 -2015 SCORE)	65.1 ± 10.4	65.3 ± 10.4	65.5 ± 10.3	64.8 ± 10.4	64.6 ± 10.4
Alcohol intake
Non drinker	15,544 (11.0%)	8,296 (10.2%)	377 (9.6%)	2,572 (12.3%)	4,299 (12.1%)
Past drinker	26,093 (18.4%)	13,489 (16.6%)	714 (18.2%)	4,421 (21.2%)	7,469 (21.0%)
<7 drinks per week	83,478 (58.9%)	49,059 (60.4%)	2,337 (59.4%)	11,789 (56.5%)	20,293 (57.0%)
7+ drinks per week	16,506 (11.7%)	10,399 (12.8%)	505 (12.8%)	2,071 (9.9%)	3,531 (9.9%)
Hormone therapy use
No	61,196 (43.2%)	44,068 (54.2%)	2,006 (51.0%)	6,868 (32.9%)	8,254 (23.2%)
Estrogen alone use	41,909 (29.6%)	5,459 (6.7%)	406 (10.3%)	12,401 (59.5%)	23,643 (66.4%)
Estrogen plus progestin use	30,567 (21.6%)	2,7874 (34.3%)	1,286 (32.7%)	481 (2.3%)	926 (2.6%)
Mixed	7,949 (5.6%)	3,842 (4.7%)	235 (6.0%)	1,103 (5.3%)	2,769 (7.8%)
Family history of cancer (yes, %)	89,580 (66.1%)	50,802 (65.2%)	2,493 (65.9%)	13,307 (66.8%)	22,978 (67.5%)
Rheumatoid arthritis ever	6,657 (4.8%)	3,374 (4.3%)	164 (4.3%)	1073 (5.3%)	2,046 (5.9%)
Lupus ever	673 (0.5%)	339 (0.4%)	20 (0.5%)	115 (0.6%)	199 (0.6%)

¹Values expressed as n (%), mean ± standard deviation. All the P values are significant (data not shown). P-value comparisons across hysterectomy categories were based on Chi-square test for categorical variables; p-values for continuous variables are based on ANOVA test.

²MET - Metabolic Equivalent of Task. HEI – Healthy eating index.

In our primary analysis, we investigated the association of hysterectomy, oophorectomy status, or combinations of hysterectomy and oophorectomy at the start of the study, as the WHI did not collect data on oophorectomy status during the follow-up period. Nevertheless, we conducted a sensitivity analysis that included updated information on hysterectomy during follow-up to evaluate the robustness of the findings.

Multivariable Cox proportional hazards regression models were used to assess the association between hysterectomy/oophorectomy status, age of hysterectomy, and NHL risk. The study conducted a Kolmogorov-type supremum test to assess the proportional hazard assumption, and the test results showed that the assumption was met. We further examined the associations between hysterectomy/oophorectomy status and risks of NHL distinct subtypes and examined whether the association between hysterectomy, oophorectomy and risk of NHL is modified by exogenous hormone use. The baseline hazard functions were stratified by study cohort (participation in OS or CTs, and different treatment assignments for all four CTs). In all multivariable models, potential confounders included all variables listed in Table 1 based on literature review and a DAG (Figure 2). Further, in a sensitivity analysis, we analyzed the exposure (hysterectomy status) as a time-varying variable using time-dependent covariate Cox model. Statistical analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC). P-values were two-sided.

Results

Among a total of 141,621 study participants, 39.9% had undergone a hysterectomy. Of those women, 45.6% stated that they had also undergone a BSO. In comparison to women without hysterectomy and without oophorectomy, those who had hysterectomy and/or oophorectomy were more likely to be older, belong to underrepresented race and Hispanic ethnicity groups, have lower levels of education, to be non-smokers, have a higher body mass index (BMI), lead a physically inactive lifestyle, consume a diet of poor quality, and to have either never drank alcohol or to be past drinkers. Additionally, they were more likely to report a family history of cancer, a history of hormone use, and a higher prevalence of rheumatoid arthritis or lupus disease (as shown in Table 1).

During an average follow-up period of 17.2 years, a total of 1,719 women developed NHL. After adjusting for potential confounders including oophorectomy, women who had undergone a hysterectomy had a 23% higher risk of NHL compared with those without a hysterectomy (HR=1.23, 95% CI: 1.05–1.44). Considering updated information on hysterectomy during the follow-up period, hysterectomy remained significantly associated with an increased NHL risk (HR=1.19, 95% CI: 1.05–1.35) (data not shown). Compared to those without hysterectomy and oophorectomy, women who had a hysterectomy between the ages of 40 to 50 years had a significantly higher risk of NHL (HR=1.36, 95% CI: 1.13–1.64) (Table 2). These findings were consistent when age at hysterectomy was analyzed as a time-varying variable.

Table 2.

Hazard ratios (HR) and 95% confidence intervals (95% CI) for the association between hysterectomy, oophorectomy and risk of Non-Hodgkin lymphoma (NHL)

	NHL Cases	Person-years	Age-adjusted HR (95% CI)	Multivariable-adjusted HR (95% CI)1
Hysterectomy 2
No	965	1,483,404	Ref	Ref
Yes	754	918,761	1.18 (1.04 1.34)	1.23 (1.05 1.44)
Oophorectomy 3
No	1192	1,770,900	Ref	Ref
One or partial	136	188,256	0.99 (0.82 1.19)	1.00 (0.83 1.21)
Bilateral oophorectomy	362	443,009	1.06 (0.91 1.23)	1.05 (0.90 1.22)
Age of hysterectomy (years) 2
No hysterectomy and no oophorectomy	924	1,418,101	Ref	Ref
<40	207	302,733	1.05 (0.88 1.25)	1.11 (0.90 1.36)
40–<50	377	417,411	1.30 (1.11 1.53)	1.36 (1.13 1.64)
50–<55	104	116,575	1.23 (0.97 1.56)	1.27 (0.98 1.63)
55+	66	82,041	0.99 (0.75 1.31)	1.02 (0.76 1.35)
P for trend among women with hysterectomy			0.85	0.82
Combinations of Hysterectomy/Oophorectomy status
None	924	1,418,101	Ref	Ref
No hysterectomy but one or part of ovaries taken out	33	59,242	0.85 (0.60 1.20)	0.85 (0.60 1.20)
No hysterectomy but bilateral oophorectomy	5	6,061	1.25 (0.52 3.01)	1.25 (0.52 3.00)
Hysterectomy without oophorectomy	268	352,799	1.16 (1.02 1.33)	1.22 (1.03 1.44)
Hysterectomy with one or part of ovaries taken out	103	129,014	1.22 (1.00 1.50)	1.31 (1.04 1.64)
Hysterectomy with bilateral oophorectomy	357	436,948	1.24 (1.10 1.40)	1.29 (1.09 1.52)
Collapsed combinations of hysterectomy and oophorectomy
No hysterectomy and no oophorectomy	924	1,418,101	Ref	Ref
No hysterectomy with oophorectomy	41	65,303	0.94 (0.68 1.28)	0.94 (0.68 1.28)
Hysterectomy without oophorectomy	268	352,799	1.16 (1.02 1.33)	1.21 (1.02 1.43)
Hysterectomy with oophorectomy	486	565,962	1.24 (1.11 1.38)	1.30 (1.11 1.51)

^1.In the multivariable-adjusted models, we adjusted potential confounders, including age at enrollment (in continuous), race (Asian, Black, White or other), Hispanic ethnicity (yes, no, missing), BMI (<25, 25.0 –<30, 30+ kg/m²), education (high school or less, some college/technical training, college or some post-college, and master or higher), diet quality (in continuous), smoking pack-years(never smoked, smoked <5, 5–<20, ≥ 20 pack-years), alcohol intake (non-drinker, past drinker, current and <7 drinks/wk, current and ≥7 drinks/wk), physical activity (<5, 5–<10, 10–<20, 20–<30, 30+ Metabolic equivalent (METs)/week, history of hormone use (no, estrogen only use, estrogen plus progestin use, mixed), family history of cancer (yes, no), rheumatoid arthritis ever (yes, no), or lupus (yes, no).

^2.This model additionally adjusted for oophorectomy.

^3.This model additionally adjusted for hysterectomy.

Oophorectomy was not independently associated with risk of NHL after adjusting for confounders including hysterectomy status (Table 2). When examining combinations of hysterectomy and oophorectomy, oophorectomy alone was not associated with risk of NHL among women without hysterectomy while women with hysterectomy had a higher risk of NHL, regardless of whether oophorectomy was performed (Table 2).

Our data show that 71% of women who had undergone a hysterectomy used unopposed estrogen. Among these women, 90% began using it after the hysterectomy procedure (data not shown). The associations between hysterectomy, age of hysterectomy, or combinations of hysterectomy and oophorectomy appear to vary by hormone use, although none of the interaction tests reached statistically significance level of 0.05. Further analysis stratified by hormone use revealed that the association between hysterectomy and NHL risk was observed specifically in women who had never used hormones (HR=1.35, 95% CI: 1.06–1.71), but not in women who had used unopposed estrogen (HR=1.09, 95% CI: 0.80, 1.47). Similarly, among women with no history of hormone use, we observed an increased risk of NHL associated with hysterectomy with and without oophorectomy compared to women without hysterectomy and oophorectomy. The highest magnitude of risk was found in women who had undergone both hysterectomy and oophorectomy (HR=1.41, 95% CI: 1.12–1.79). However, we did not detect that the risk of NHL with hysterectomy significantly varied by oophorectomy status. Furthermore, when examining associations by hysterectomy age and hormone use, this association was confined to women who had undergone hysterectomy before the age of 55 with no history of hormone use (Table 3).

Table 3.

Hazard ratios (HR) and 95% confidence intervals (95% CI) for the association between hysterectomy and risk of Non-Hodgkin lymphoma (NHL) by hormone use

	Never hormone users			Unopposed estrogen users			Interaction test
	NHL Cases/Person-Years	Age-adjusted HR (95% CI)	Multivariable-adjusted HR (95% CI)1	NHL Cases/Person-Years	Age-adjusted HR (95% CI)	Multivariable-adjusted HR (95% CI)1
Hysterectomy 2							0.13
No	498/767,106	Ref	Ref	75/92,416	Ref	Ref
Yes	194/235,436	1.25 (1.01 1.54)	1.35 (1.06 1.71)	483/615,271	1.09 (0.83 1.43)	1.09 (0.81 1.47)
Collapsed combinations of hysterectomy and oophorectomy							0.14
No hysterectomy and no oophorectomy	479/734,672	Ref	Ref	67/86,353	Ref	Ref
No hysterectomy with oophorectomy	19/32,434	0.88 (0.56 1.39)	0.88 (0.55 1.39)	8/6,063	1.71 (0.82 3.55)	1.66 (0.79 3.46)
Hysterectomy without oophorectomy	84/110,142	1.21 (0.96 1.52)	1.30 (1.00 1.68)	170/213,954	1.17 (0.88 1.55)	1.15 (0.84 1.58)
Hysterectomy with oophorectomy	110/125,294	1.33 (1.08 1.63)	1.41 (1.12 1.79)	313/401,316	1.13 (0.86 1.48)	1.13 (0.84 1.52)
Age of hysterectomy (years) 2							0.06
No hysterectomy and no oophorectomy	479/734,672	Ref	Ref	67/86,353	Ref	Ref
<40	69/87,971	1.31 (1.02 1.69)	1.43 (1.08 1.89)	120/203,640	0.91 (0.67 1.24)	0.92 (0.66 1.29)
40–<50	84/101,739	1.27 (1.01 1.60)	1.35 (1.04 1.74)	262/289,728	1.30 (0.99 1.71)	1.29 (0.95 1.75)
50–<55	27/24,140	1.59 (1.08 2.35)	1.63 (1.09 2.43)	61/78,376	1.07 (0.75 1.51)	1.04 (0.72 1.51)
55+	14/21,586	0.85 (0.50 1.44)	0.88 (0.51 1.51)	40/43,526	1.10 (0.75 1.63)	1.07 (0.70 1.61)

^1.In the multivariable-adjusted models, we adjusted potential confounders, including age at enrollment (in continuous), race (Asian, Black, White or other), Hispanic ethnicity (yes, no, missing), BMI (<25, 25.0 –<30, 30+ kg/m²), education (high school or less, some college/technical training, college or some post-college, and master or higher), diet quality (in continuous), smoking pack-years(never smoked, smoked <5, 5–<20, ≥ 20 pack-years), alcohol intake (non-drinker, past drinker, current and <7 drinks/wk, current and ≥7 drinks/wk), physical activity (<5, 5–<10, 10–<20, 20–<30, 30+ Metabolic equivalent (METs)/week, family history of cancer (yes, no), rheumatoid arthritis ever (yes, no), or lupus yes, no).

^2.This model additionally adjusted for oophorectomy.

When investigating the relationship between hysterectomy and different subtypes of NHL, it was noted that hormone use played a significant role in modifying the association between hysterectomy, combinations of hysterectomy and oophorectomy, and the risk of diffuse large B-cell lymphoma (DLBCL) (p for interaction=0.01, 0.03, respectively). Among women who had never used hormones, hysterectomy was associated with an increased risk of DLBCL, whereas this association was not observed among women who used unopposed estrogen. On the other hand, the study did not find significant modification by hormone use in the association between hysterectomy and the risk of FL (p for interaction=0.84), or risk of CLL/SLL (p for interaction=0.87). In addition, an elevated risk of FL or CLL/SLL was observed in relation to hysterectomy, although the association did not reach statistical significance. Similar results were observed when combinations of hysterectomy and oophorectomy were analyzed (Table 4).

Table 4.

Hazard ratios (HR) and 95% confidence intervals (95% CI) for the association between hysterectomy, oophorectomy and risk of Non-Hodgkin’s lymphoma (NHL) subtypes¹

	Diffuse large B-cell lymphoma (DLBCL)		Follicular lymphoma (FL)		Chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL)
	NHL Cases	Multivariable-adjusted HR (95% CI)	NHL Cases	Multivariable-adjusted HR (95% CI)	NHL Cases	Multivariable-adjusted HR (95% CI)
Hysterectomy 2
No	278	Ref	174	Ref	76	Ref
Yes	207	1.20 (0.89 1.62)	151	1.39 (0.95 2.02)	58	1.54 (0.87 2.72)
Never hormone users 2
Hysterectomy (No)	135	Ref	83	Ref	36	Ref
Hysterectomy (Yes)	65	1.48 (0.96 2.28)	29	1.22 (0.66 2.25)	18	1.41 (0.62 3.23)
Unopposed estrogen users 2
Hysterectomy (No)	27	Ref	12	Ref	3	Ref
Hysterectomy (Yes)	124	0.83 (0.50 1.38)	103	1.27 (0.63 2.55)	33	-
Interaction between hysterectomy and hormone use 2	0.01		0.84		0.87
Collapsed combinations of hysterectomy and oophorectomy
No hysterectomy and no oophorectomy	267	Ref	169	Ref	74	Ref
Hysterectomy without oophorectomy	86	1.21 (0.89 1.65)	54	1.33 (0.90 1.97)	17	1.25 (0.66 2.37)
Hysterectomy with oophorectomy	121	1.02 (0.76 1.37)	97	1.40 (0.98 2.00)	41	1.80 (1.06 3.08)
Never hormone users
No hysterectomy and no oophorectomy	129	Ref	82	Ref	35	Ref
Hysterectomy without oophorectomy	29	1.51 (0.96 2.38)	12	0.99 (0.50 1.97)	7	1.21 (0.48 3.05)
Hysterectomy with oophorectomy	36	1.60 (1.05 2.44)	17	1.21 (0.66 2.19)	11	1.48 (0.66 3.32)
Unopposed estrogen users
No hysterectomy and no oophorectomy	24	Ref	10	Ref	3	Ref
Hysterectomy without oophorectomy	55	0.91 (0.53 1.54)	37	1.47 (0.69 3.13)	9	_
Hysterectomy with oophorectomy	69	0.61 (0.36 1.01)	66	1.39 (0.68 2.85)	24	_
Interaction between combinations of hysterectomy and oophorectomy and hormone use	0.03		0.28		0.99

^1.In the multivariable-adjusted models, we adjusted potential confounders, including age at enrollment (in continuous), race (Asian, Black, White or other), Hispanic ethnicity (yes, no, missing), BMI (<25, 25.0 –<30, 30+ kg/m²), education (high school or less, some college/technical training, college or some post-college, and master or higher), diet quality (in continuous), smoking pack-years(never smoked, smoked <5, 5–<20, ≥ 20 pack-years), alcohol intake (non-drinker, past drinker, current and <7 drinks/wk, current and ≥7 drinks/wk), physical activity (<5, 5–<10, 10–<20, 20–<30, 30+ Metabolic equivalent (METs)/week, family history of cancer (yes, no), rheumatoid arthritis ever (yes, no), or lupus yes, no).

^2.Oophorectomy was additionally adjusted for in these analyses.

Discussion

In this large prospective study, we observed that women who underwent a hysterectomy had an approximately 23% increased risk of NHL compared to women without hysterectomy, regardless of oophorectomy status. The association between hysterectomy and NHL risk was particularly pronounced in women who had the procedure younger than 55 and had no history of hormone use. Furthermore, our analysis indicated that hormone use significantly modified the association of hysterectomy with NHL for DLBCL subtype.

In line with our results, other studies have observed increased NHL risk in association with hysterectomy.^{15, 19} There are several possible explanations for the increased risk of NHL associated with hysterectomy. First, it may be due to hormonal changes. Hysterectomy involves the removal of the uterus and often involves the removal of the ovaries as well. This procedure can result in relative estrogen deficit. These hormonal changes have the potential to impact the immune system, which may play a role in the development of NHL.³⁰ For example, studies have shown that relatively higher estrogen levels in premenopausal women suppress the release of pro-inflammatory cytokines, such as interleukins IL-6 and IL-8, as well as tumor necrosis factor α (TNF-α).^{31, 32} However, we did not observe the risk of NHL with hysterectomy significantly varied by oophorectomy status. Second, findings may be explained by surgical trauma and inflammation. The surgical procedure itself, including tissue trauma and inflammation, may trigger an immune response in the body.³³ Third, the association may be due to shared risk factors. It is possible that women who undergo hysterectomy may have other underlying risk factors for NHL that are not directly related to the surgery itself. For example, certain genetic factors or lifestyle choices may be more prevalent in women undergoing hysterectomy, which could contribute to the observed increased risk. Despite our efforts to control for potential confounding variables, there may be other factors that were not considered or adequately accounted for. These factors could potentially explain the observed association between hysterectomy and NHL.

Our study revealed that the association between hysterectomy and risk of NHL was modified by exogenous unopposed estrogen use, especially for DLBCL subtype. Specifically, we observed that among women who had never used hormones, hysterectomy was linked to an increased risk of DLBCL. However, this association was not observed among women who used unopposed estrogen. Although the molecular mechanisms by which hormones impact the development of NHL remains elusive, it is hypothesized that the influence of estrogen on the body’s immune response could be one potential mechanism. Estrogen receptors have been found in normal B and T lymphocytes, bone marrow, as well as in cell lines of leukemia and lymphoma.¹¹ Studies indicate that estrogen, specifically through its binding to ER-β receptors, may have an anti-tumor effect in lymphoma.^{34, 35}

It remains unclear why hormone use played a significant role in modifying the association with hysterectomy for DLBCL subtypes but not for other examined NHL subtypes. One possibility is that NHL subtypes exhibit differences in hormone receptor expression patterns, which could explain the observed discrepancies in association with hormone use.³⁶ For instance, studies have shown significantly higher levels of ER-β expression in DLBCL compared to normal B cells.^{36, 37} Although results of studies of menopausal hormone use and risk of NHL are mixed,^{12, 14, 16, 18} results from several observational studies found reduced risk of DLBCL among estrogen-only users^{12, 13, 16} but higher risk associated with FL.^{12, 13} Additionally, the lack of significant effect modification for CLL/SLL may be due to the limited sample size, which could have hindered our ability to detect a meaningful association. Therefore, further research is necessary to comprehensively understand the underlying mechanisms contributing to DLBCL development and its potential interactions with hormone use.

We observed that the association between age at hysterectomy and risk of NHL was significant in women who had the procedure at age younger than 55 years and had no history of hormone use. The absence of an association between hysterectomy at an older age and the risk of NHL may indicate that the impact of hormonal factors on NHL development is diminished or less pronounced in older individuals compared to younger ones. Also, it is possible that the contribution of hysterectomy to NHL development in older women may be overshadowed by the influence of other established factors, resulting in no observable association.

The present study has several strengths, including its prospective design, large sample size, long follow-up duration, comprehensive information on potential confounders, and adjudicated outcomes and assessment of subtypes of outcomes. However, it is important to acknowledge several limitations as well. First, misclassification of hysterectomy and oophorectomy status based on self-reported information is possible. However, any misclassification is likely to have been non-differential, meaning that it would have biased the estimates toward the null. Second, there is a potential for detection bias due to more intensive medical surveillance in recent years following hysterectomy. However, our findings indicate that the association between hysterectomy and NHL risk did not substantially change with time since hysterectomy, suggesting that detection bias is not a likely explanation. Third, the study lacked information on specific types of hysterectomy (such as vaginal or abdominal hysterectomy, laparoscopic procedures) and the underlying indications for hysterectomy (such as uterine fibroids, endometriosis, ovarian tumors, or uterine prolapse). This limited information prevents a more detailed analysis of how these factors may influence the association between hysterectomy and NHL risk. Fourth, only 77% were re-consented for extension studies after completion of the main study, however, there is no data showing differential dropouts in relation to our exposure and outcome status. An additional limitation of the study is the possibility of residual confounding from unmeasured factors (such as radiation, chemical exposures) that were not accounted for in the analysis. Finally, the study population only consisted of postmenopausal women, and the findings may not necessarily apply to younger women. It is important to conduct similar studies that include premenopausal women to assess whether the findings differ in that population.

To summarize, our large prospective study found that women with a history of hysterectomy had an approximately 26% higher risk of NHL, regardless of oophorectomy status or type of oophorectomy. The association between hysterectomy and NHL risk was particularly evident in women who had never used hormones, especially for the DLBCL subtype. Considering that hysterectomy is a common surgical procedure among women in the United States, this finding warrants further investigation. Future studies incorporating detailed information on the types and indications of hysterectomy and a meta-analysis utilizing prior data may help to shed light on this association and deepen our understanding of the mechanisms underlying DLBCL development and its potential interactions with hormone use.

Novelty and impact:

Hysterectomy is a common surgical procedure among women. Our study revealed that women who underwent a hysterectomy had 23% increased risk of NHL compared to women without hysterectomy, regardless of oophorectomy status. The association was particularly pronounced in women who had no history of hormone use, especially for the diffuse large B-cell lymphoma subtype. Our findings, if confirmed, could require healthcare providers to discuss this potential risk with patients considering a hysterectomy.

Abbreviations:

BMI

body mass index

BSO

bilateral salpingo-oophorectomy

CL

confidence interval

CLL/SLL

Chronic lymphocytic leukemia/small lymphocytic lymphoma

CT

clinical trial

DLBXL

diffuse large B-cell lymphoma

ER

estrogen receptor

FL

follicular lymphoma

HEI

Healthy Eating Index

HR

hazard ratio

IL

interleukin

MET

metabolic equivalent task

NHL

non-Hodgkin’s lymphoma

OS

observational study

TNF-α

tumor necrosis factor α

WHI

Women’s Health Initiative

Data Availability Statement:

The data that support the findings of this study are available from the Women’s Health Initiative (WHI). Data can be accessed upon reasonable requests made to WHI. Further information is available from the corresponding author upon request.