Abstract
Objective
Prior studies suggest an increased risk of cervical cancer among women with systemic lupus erythematosus (SLE). However, the relationship with immunosuppressive drugs (ISDs) is not well studied in U.S. nationwide cohorts. We compared the risk of high-grade cervical dysplasia and cervical cancer among women with SLE who started ISDs versus hydroxychloroquine (HCQ).
Methods
We identified SLE patients initiating ISD or HCQ using claims data from two US commercial health plans and Medicaid (2000–2012). We used a validated claims-based algorithm to identify high-grade cervical dysplasia or cervical cancer. To account for potential confounders including demographic factors, comorbidities, medication use, HPV vaccination status, and health care utilization, ISD and HCQ initiators were 1:1 matched on the propensity score (PS). We used inverse variance-weighted, fixed effect models to pool hazard ratios (HR) from the PS-matched Medicaid and commercial cohorts.
Results
We included 2,451 matched pairs of ISD and HCQ new users in the commercial cohort and 7,690 matched pairs in Medicaid. In the commercial cohort, there were 14 cases of cervical dysplasia or cervical cancer among ISD users and 5 cases among HCQ users (HR 2.47, 95% CI 0.89–6.85, HCQ=ref). In Medicaid, there were 46 cases among ISD users and 29 cases in HCQ users (HR 1.24, 95% CI 0.78–1.98, HCQ=ref). The pooled HR of ISD was 1.40 (95% CI 0.92–2.12).
Conclusion
Among women with SLE, ISDs may be associated with a greater, albeit not statistically significant, risk of high-grade cervical dysplasia and cervical cancer compared to patients receiving HCQ alone.
Introduction
In the U.S. there are nearly 13,000 new cases of invasive cervical cancer and over 4,000 related deaths each year.1 Human papillomavirus (HPV) is the most common sexually transmitted disease among females in the U.S. and the main cause of high-grade cervical dysplasia and cervical cancer.2 High-grade cervical dysplasia refers to cervical lesions which are classified as cervical intraepithelial neoplasia 2 and 3 and because of the increased risk of progression to cervical cancer compared to lower grade lesions, they require aggressive treatment.3 While cancer rates are expected to decrease with increased use of efficacious HPV vaccines, vaccine uptake in the U.S. remains low.4
A number of studies suggest that patients with systemic inflammatory diseases and systemic lupus erythematosus (SLE) in particular, may be at increased risk for persistent HPV infection, high-grade cervical dysplasia, and cervical cancer.5 This heightened vulnerability may be due to an immunocompromised state from the disease itself, the immunosuppressive drugs (ISDs) used, or a combination of both resulting in a diminished ability to clear the HPV infection.6 Few studies to date have assessed the relationship between ISD use and cervical cancer among patients with systemic inflammatory diseases. To our knowledge, there are no U.S.-based nationwide longitudinal studies that assess the potential relationship between ISD use and incidence of high-grade cervical dysplasia and cervical cancer among patients with SLE. We hypothesized that ISD use will be associated with increased incidence of high-grade cervical dysplasia and cancer when compared to SLE patients receiving hydroxychloroquine alone.
Material and Methods
Data Source
To conduct this cohort study, we used administrative claims data from two U.S. commercial health plans, HealthCore (January 1, 2001–June 30, 2008), United HealthCare (January 1, 2003–December 31, 2012), and from Medicaid (January 1, 2000–December 31, 2010). The commercial health plans primarily provide insurance for working adults and their family members nationwide in the U.S. These databases have been described in detail in a prior study.5 Medicaid supports low-income individuals and is the largest public insurance in the U.S. Medicaid data, including demographic information and billing claims for beneficiaries in 49 states and Washington, DC, are part of the Medicaid Analytic eXtract database. Data from both commercial health plans and from Medicaid were de-identified and therefore patient informed consent was not required. The study was approved by the Brigham and Women’s Hospital Institutional Review Board.
Study Population
We identified females aged ≥18 years with SLE, defined by ≥2 International Classification of Diseases, Ninth Revision (ICD-9) codes (710.X) on two occasions, separated by ≥7 days. Within this cohort, we selected one group of individuals who newly started an ISD and one group who newly started HCQ. ISDs included methotrexate, azathioprine, mycophenolate mofetil, cyclophosphamide (oral or intravenous), tacrolimus, abatacept, rituximab, cyclosporine or belimumab. We defined the start of follow-up (the index date) as the date of first dispensing of ISD or HCQ. The ISD group was required to have no use of any ISD in the prior 365 days of continuous enrollment. The HCQ group was required to have no use of HCQ or ISDs in the prior 365 days of continuous enrollment. Exclusion criteria included a history of hysterectomy, organ or bone marrow transplant, receipt of chemotherapy, HIV/AIDS, or residence in a nursing home during the 365 days prior to the index date. Patients were followed from the index date to the earliest occurrence of discontinuation of HCQ or start of an ISD for the HCQ group, discontinuation of ISD for the ISD group, development of high-grade cervical dysplasia or cervical cancer, disenrollment from the health plan, end of the database period, or death.
Outcome of Interest
Our primary outcome was high-grade cervical dysplasia or cervical cancer assessed beginning at the index date. To define this we used a validated claims-based algorithm of ≥2 ICD-9 codes (795.04, 622.12, 233.1, 180.1) and ≥1 procedure code for related gynecologic procedures or treatments (CPT codes for colposcopy, cervical biopsy, cryotherapy of the cervix, electro or thermal cautery of the cervix, endocervical curettage, laser ablation, cervical conization, loop excision, cervicectomy, or excision of cervical stump) occurring ≥30 days after the ICD-9 code (positive predictive value of ≥81%) (Table 1).7 We also examined the rate of procedures related to cervical dysplasia including Papanicolaou test, colposcopy, cervical biopsy and cervical excision surgery during the follow-up period.
Table 1.
Codes for identification of baseline covariates and outcome
| Baseline covariates* or outcome | ICD-9, medication use or CPT code |
|---|---|
| Diabetes | ICD-9 code 250.xx or anti-diabetic medication use |
| Chronic kidney disease | ICD-9 codes 580–588, v45.1x |
| Liver disease | ICD-9 codes 571–573, 070.x |
| Alcoholism | ICD-9 codes 305.0x, 291.9, 291.89, 303.0, 291.0, 291.3 571.0, 571.1, 571.2, 571.3 |
| Smoking | ICD-9 codes 305.1x, 649.0x, 989.84 or medication use for smoking cessation or CPT codes 99406, 99407, S9075, S9453 |
| Substance abuse | ICD-9 codes 304.x, 305.2x–305.7x, or CPT codes 99408, 99409 |
| Sexually transmitted infection | ICD-9 codes 054.1x, 091.x–099.x, 078.11, 078.88–078.89, 079.4x, 079.9x, 131.x, V69.2 |
| HPV vaccination | ICD-9 codes V05.8 or CPT codes 90649, 90650 |
| Mammography | CPT codes 76090, 76091, 76092 |
| Papanicolaou test | CPT codes 88142, 88164, 88175, 88141 |
| HPV DNA test | CPT codes 87624, 87621 |
|
Outcome: Cervical cancer or high- grade cervical dysplasia (defined by 2 ICD-9 codes and 1 CPT code for procedure within 30 days) |
ICD-9 codes 180.1, 795.04, 622.12, 233.1; CPT codes 57420, 57421, 57452, 57454, 57455, 57445, 57460, 57461, 57500, 57505, 57511, 57510, 57513, 57520, 57522, 57530, 57531, 57540, 57550 |
Comorbidity score and likelihood of sexual activity were determined using previous defined algorithms of ICD-9 and CPT codes, contraceptive use determined by medication or long-acting reversible contraceptive use, lipid screening determined by laboratory tests ordered for cholesterol, HDL or LDL
Covariates
We measured all covariates during the 365 days prior to the index date (Table 1). These included age, comorbidities (diabetes, chronic kidney disease, liver disease, and a comorbidity score8), cervical cancer-related factors (alcoholism, smoking, substance use, likelihood of sexual activity9, sexually transmitted infection, oral and non-oral contraceptive use, receipt of ≥1 HPV vaccine), health care utilization (outpatient visits, rheumatology visits, obstetrics/gynecology visits, inpatient admissions, emergency department visits), preventive care (mammography, Papanicolaou test, HPV DNA test, lipid screening), and cumulative dose of systemic corticosteroids during the baseline period. The cumulative steroid dose was calculated based on the number of prednisone-equivalent milligrams prescribed for the 365 days prior to the index date.
Statistical Analyses
We first compared differences in baseline characteristics between HCQ and ISD new users in Medicaid and in the commercial health plans. We then used multivariable logistic regression models to calculate propensity scores for receipt of ISD vs. HCQ accounting for all of the aforementioned baseline covariates including age, comorbidities, cervical cancer-related factors, healthcare utilization, preventive care and corticosteroid use. We used nearest neighbor matching within a caliper of 0.05 on the propensity score to match ISD users to HCQ users at a fixed ratio of 1:1.10, 11 The c-statistic for the commercial plan beneficiaries was 0.72 and for Medicaid beneficiaries was 0.66. We determined incidence rates (IR) and 95% confidence intervals (CIs) of high-grade cervical dysplasia and cervical cancer in the propensity score matched cohorts and Cox proportional hazards models were used to estimate the hazard ratio (HR) with 95% CI of these outcomes associated with use of ISD vs. HCQ. We tested the proportional hazards assumption (exposure multiplied times log of follow-up time) and it was not violated in any of our models. In secondary analyses, we compared rates of gynecologic procedures related to cervical dysplasia between the ISD and HCQ groups. We also examined the IR and 95% CIs of high-grade cervical dysplasia and cervical cancer by individual immunosuppressive agent in both the Medicaid and commercial health plan datasets. Due to anticipated differences in the patient populations enrolled in Medicaid compared the commercial health plans, we conducted all of our analyses, including propensity score development, separately in Medicaid and in the combined commercial health plan databases. We then pooled the database-specific HRs using an inverse variance-weighted fixed effects model, in order to augment our statistical power to detect meaningful differences in outcomes between the ISD and HCQ groups. All analyses were conducted using SAS 9.3 (Cary, NC) and STATA 14.
Results
In the commercial health plans, 36,691 females met our criteria for SLE and 9,731 were new users of ISD or HCQ. In Medicaid, we identified 67,851 females with SLE, 24,877 of whom were new users of ISD or HCQ. After applying our additional inclusion and exclusion criteria, we identified 7,223 females in the commercial health plans and 19,861 in Medicaid (Figure 1). Following 1:1 propensity score matching, we included 2,451 matched pairs of ISD and HCQ initiators in the commercial health plans and 7,690 matched pairs in Medicaid (Table 2). The mean ± SD follow-up time beginning at the index date of new medication use was 1.1 ± 1.4 years in the commercial health plans and 1.5 ± 1.9 years in Medicaid.
Figure 1.
Cohort Selection Flow
Table 2.
Baseline characteristics of hydroxychloroquine (HCQ) or immunosuppressive drug (ISD) initiators with SLE
| Database | Commercial Insurance Plans | Medicaid Insurance | ||
|---|---|---|---|---|
| Medication New Users | HCQ (n=2,451) | ISD (n=2,451) | HCQ (n=7,690) | ISD (n=7,690) |
| Age –mean ± SD | 45.7 ± 12.1 | 45.2 ± 12.1 | 39.2 ± 12.0 | 39.2 ± 12.4 |
| HPV-associated factors – N (%) | ||||
| Sexually active | 1920 (78.3) | 1943 (79.3) | 5560 (72.3) | 5569 (72.4) |
| Sexually transmitted disease | 150 (6.1) | 141 (5.8) | 729 (9.5) | 743 (9.7) |
| Alcoholism | 9 (0.4) | 10 (0.4) | 50 (0.7) | 29 (0.6) |
| Substance abuse | 20 (0.8) | 23 (0.9) | 163 (2.1) | 162 (2.1) |
| Smoking | 247 (10.1) | 265 (10.8) | 976 (12.7) | 976 (12.7) |
| Oral contraceptives | 225 (9.2) | 232 (9.5) | 566 (7.4) | 588 (7.7) |
| Non-oral contraceptives | 41 (1.7) | 39 (1.6) | 287 (3.7) | 279 (3.6) |
| Receipt of HPV vaccine | 15 (0.6) | 15 (0.6) | 33 (0.4) | 32 (0.4) |
| Abnormal Papanicolaou test | 130 (5.3) | 142 (5.8) | 409 (5.3) | 420 (5.5) |
| Comorbidities – N (%) | ||||
| Chronic kidney disease | 270 (11.0) | 253 (10.3) | 1156 (15.0) | 1186 (15.4) |
| Liver disease | 169 (6.9) | 166 (6.8) | 563 (7.3) | 542 (7.1) |
| Diabetes mellitus | 270 (11.0) | 252 (10.3) | 1106 (14.4) | 1159 (15.1) |
| Mean comorbidity score ± SD* | 0.8 ± 1.6 | 0.8 ± 1.5 | 1.1 ± 1.8 | 1.1 ± 1.7 |
| Medications | ||||
| Mean number of prescriptions ± SD | 12.2 ± 7.4 | 12.2 ± 7.0 | 16.4 ± 9.2 | 16.5 ± 8.6 |
| Corticosteroid use – N (%) | 1006 (41.0) | 1004 (41.0) | 4252 (55.3) | 4171 (54.2) |
| Preventive Care – N (%) | ||||
| Mammogram | 890 (36.3) | 858 (35.0) | 1367 (17.8) | 1385 (18.0) |
| Papanicolaou test | 1065 (43.5) | 1058 (43.2) | 2285 (29.7) | 2310 (30.0) |
| HPV DNA ordered | 218 (8.9) | 215 (8.8) | 346 (4.5) | 356 (4.5) |
| Colonoscopy | 243 (9.9) | 234 (9.6) | 453 (5.9) | 456 (5.9) |
| Healthcare utilization- mean ± SD | ||||
| Outpatient visits | 12.5 ± 8.3 | 12.7 ± 8.9 | 10.8 ± 9.3 | 10.9 ± 8.6 |
| Hospitalizations | 0.6 ± 2.5 | 0.6 ± 1.9 | 0.7 ± 1.4 | 0.7 ± 1.4 |
| Emergency Department visits | 0.7 ± 3.2 | 0.8 ± 3.4 | 2.2 ± 3.9 | 2.1 ± 3.8 |
The range of the comorbidity score is −2 to 26.
During the baseline period, Medicaid enrollees overall were younger, had more comorbidities and Emergency Department visits, and received more prescriptions and less preventive care services compared to commercial plan enrollees (Table 2). More the 75% of women in both cohorts were classified as sexually active and very few received the HPV vaccine (0.6% in commercial plans and 0.4% in Medicaid) during the baseline period.
Among commercial plan enrollees with SLE, there were 14 cases of high-grade cervical dysplasia or cervical cancer associated with ISD use (IR per 4.70 per 1,000 person-years, 95% CI 2.78–7.94) and 5 cases associated with HCQ use (IR 1.89 per 1,000 person-years, 95% CI 0.79–4.54) (Table 3). The HR was 2.47 (95% CI 0.89–6.85), with HCQ as the reference. Of the 14 cases, 5 occurred among azathioprine users (IR 8.32 per 1,000 person-years, 95% CI 2.46–20.00), 5 among methotrexate users (IR 4.25 per 1,000 person-years, 95% CI 1.77–10.20), and 4 among mycophenolate mofetil users (IR 8.72 per 1,000 person-years, 95% CI 3.27–23.22).
Table 3.
Risk of high-grade cervical dysplasia or cervical cancer among SLE patients associated with initiation of immunosuppressive drugs versus hydroxychloroquine in a propensity-score matched analysis
| Immunosuppressive Drugs | Hydroxychloroquine | |||||||
|---|---|---|---|---|---|---|---|---|
| Cases | Person- years |
Incidence Rate* (95% CI) |
Hazard ratio (95% CI) |
Cases | Person- years |
Incidence Rate* (95% CI) |
Hazard ratio (95% CI) |
|
|
Commercial Health Plan |
14 | 2976 | 4.70 (2.78–7.94) |
2.47 (0.89–6.85) |
5 | 2646 | 1.89 (0.79–4.54) |
Ref. |
| Medicaid | 46 | 13,043 | 3.53 (2.64–4.71) |
1.24 (0.78–1.98) |
29 | 10,772 | 2.69 (1.87–3.87) |
Ref. |
| Pooled Analysis | 1.40 (0.92–2.12) |
Ref. | ||||||
Incidence rate per 1,000 person-years
In Medicaid, there were 46 cases of high-grade cervical dysplasia associated with ISD use (IR 3.53 per 1,000 person-years, 95% CI 2.64–4.71) and 29 associated with HCQ use (IR 2.69 per 1,000 person-years, 1.87–3.87). Comparing ISD to HCQ, the HR associated with ISD use was 1.24 (95% CI 0.78–1.98). Of the 46 cases, 16 occurred among mycophenolate mofetil users (IR 7.04 per 1,000 person-years, 95% CI 4.32–11.50), 12 occurred among methotrexate users (IR 2.55 per 1,000 person-years, 95% CI 1.45–4.50) and 11 occurred among azathioprine users (2.62 per 1,000 person-years, 95% CI 1.45–4.74). Rates for other ISDs are suppressed due to fewer than 11 outcomes, in accordance with Centers for Medicare and Medicaid Services policy.
In the pooled analyses across health plans, the HR associated with ISD use was 1.40 (95% CI 0.92–2.12). We also examined use of gynecologic procedures in the matched cohorts (Table 4). Among commercial health plan enrollees, we observed a slightly higher incidence rate of Papanicolaou tests and gynecologic procedures overall among ISD initiators compared to HCQ initiators. Among Medicaid enrollees, we did not observe any significant differences in gynecologic procedure use associated with ISD or HCQ use.
Table 4.
Incidence rates (IR) and incidence rate ratios (IRR) of gynecologic visits and procedures associated with use of immunosuppressive drugs (ISD) versus hydroxychloroquine (HCQ) in a propensity-score matched analysis
| Immunosuppressive Drugs | Hydroxychloroquine | |||||||
|---|---|---|---|---|---|---|---|---|
| Number | Person- years |
IR per 1,000 person-years (95% CI) |
IRR (95% CI) |
Number | Person- years |
IR per 1,000 person-years (95% CI) |
IRR (95% CI) |
|
| Commercial Insurance Plans | ||||||||
| Outpatientgynecology visits | 2912 | 3000 | 970.82 (936.20– 10006.70) |
0.93 (0.88–0.98) |
2773 | 2655 | 1044.52 (1006.36–1084.13) |
Ref. |
| Papanicolaou tests | 566 | 3000 | 188.70 (173.77–204.9) |
1.15 (1.01–1.27) |
435 | 2655 | 163.85 (149.16–180.0) |
Ref. |
| Colposcopy | 90 | 3000 | 30.0 (24.40–36.89) |
0.93 (0.69–1.25) |
86 | 2655 | 32.39 (26.22–40.02) |
Ref. |
| Any gynecologic procedure | 679 | 3000 | 226.37 (209.97–244.05) |
1.13 (1.01–1.27) |
521 | 2655 | 200.02 (183.71–217.77) |
Ref. |
| Medicaid Insurance* | ||||||||
| Papanicolaou tests | 471 | 3621 | 130.09 (118.86–142.39) |
1.00 (0.94–1.20) |
411 | 3164 | 129.91 (117.9–143.10) |
Ref. |
| Colposcopy | 121 | 3621 | 33.42 (27.97–39.94) |
1.31 (0.99–1.74) |
81 | 3164 | 25.60 (20.59–31.83) |
Ref. |
| Any gynecologic procedure | 618 | 3621 | 170.70 (157.75–184.70) |
1.10 (0.94–1.20) |
507 | 3164 | 160.25 (146.89–174.83) |
Ref. |
Number of outpatient gynecologic visits is not available in Medicaid data
Discussion
In our analyses using U.S. nationwide commercial and public insurance databases, we did not observe a statistically significant difference in rates of high-grade cervical dysplasia and cervical cancer associated with ISD use. However, in both cohorts, and in our pooled analysis, we did note a trend towards higher rates among ISD initiators compared to HCQ initiators. We did not find statistically significant differences in incidence rates of high-grade cervical dysplasia or cervical cancer by individual ISD, however the small number of outcomes limited this aspect of our analysis. We found slightly higher rates of gynecologic procedures among ISD initiators compared to HCQ initiators with commercial insurance, but no difference among those with Medicaid.
To date, there are only a few studies that specifically examined the association between ISD use and cervical cancer in patients with systemic autoimmune diseases. One longitudinal, population-based cohort study in Denmark compared the risk of cervical cancer among patients with autoimmune diseases to the general population.12 The authors found similar rates of cervical cancer in patients with autoimmune disease and in the general population and this held true for those who received ISDs. However, when azathioprine was examined independently, while there was no statistically significant increased risk among users overall compared to non-users, there was a 2.2 times higher risk (95% CI 1.2–3.9) among those receiving the highest cumulative dose compared to non-users. One cross-sectional study among patients with SLE (n=34) and rheumatoid arthritis (n=43) in Mexico, found an association between methotrexate use and HPV positivity.13 A small study (n=40) among patients with inflammatory bowel disease (IBD) compared to healthy controls, found an increased risk of abnormal Papanicolaou tests and of high-grade cervical dysplasia among IBD patients and this was most pronounced among those receiving ISDs.14
Among SLE patients, prior studies suggested the possibility of an increased risk of cervical dysplasia conferred by ISD use. One small study compared women with SLE to age-matched healthy controls and found an increased rate of cervical atypia (19/80 in SLE patients and 9/80 in controls).15 The majority (18 of the 19 SLE patients) reported treatment with azathioprine which raised the potential question of associated risk. A longitudinal study of 61 SLE patients evaluated the three-year incidence of cervical intraepithelial neoplasia.16 The authors identified six cases, two in patients who received cyclophosphamide and prednisone and four who received cyclophosphamide, azathioprine and prednisone. With the small numbers and the likelihood of confounding by indication, it is challenging to extrapolate these findings to the general population of SLE patients receiving ISD. One larger study included 171 SLE patients, 87 of whom received ISDs, and these patients were compared to healthy controls.17 A higher risk of cervical dysplasia was seen among the SLE patients compared to the controls, and SLE patients with at least one year of ISD use had a higher prevalence of both low and high-grade intraepithelial lesions compared with those who did not receive these medications. One multi-centered academic cohort study of 1,015 SLE patients, 41 percent of whom received ISDs, found a trend towards increased risk of an abnormal Papanicolaou test in their adjusted analyses (Odds ratio 1.6, 95% 1.0–2.7), which included 74 abnormal tests following SLE diagnosis.18
Our study has a number of strengths. First, our study cohorts are population-based and representative of both commercial health plan members and Medicaid beneficiaries in the U.S. We were able to assess the comparative risk of high-grade cervical dysplasia, a surrogate endpoint for cervical cancer, and cervical cancer in a large U.S.-based longitudinal cohort of over 20,000 SLE patients initiating and actively receiving either an ISD or HCQ. Since individuals enrolled in commercial health plans tend to be different from those enrolled in Medicaid, we examined this question in each population separately first and subsequently pooled the HRs. A number of prior studies compared the risk of cervical dysplasia among SLE patients receiving ISD to healthy controls, or to SLE patients on a variety of other medications, or no medications at all. We used the new user design with active comparators (i.e. ISD initiators vs. HCQ initiators) to avoid immortal time bias as well as the issue with depletion of the susceptibles and to ensure that all of the patients studied were continuously on treatment with either ISD or HCQ during the follow-up period. To minimize the potential for differences in factors including disease severity among ISD and HCQ initiators (i.e., confounding by indication) and improve statistical efficiency, our study cohorts were matched on the propensity score. In our propensity score models, we further controlled for potential differences in baseline preventative health services including cervical cancer screening between ISD and HCQ initiators. In addition, we used a validated claims-based algorithm, shown to have a positive predictive value of greater than 80 percent, to identify our outcomes of high-grade cervical dysplasia and cervical cancer.7
There are several limitations to this study. While this is the largest U.S.-based longitudinal study to date to examine the association of ISD with high-grade cervical dysplasia and cervical cancer among SLE patients, our outcome of interest was uncommon. Therefore, our power to detect a clinically or statistically significant difference between the groups is limited. However, other strategies that could have further increased our sample size and our number of outcomes, such as including prevalent users rather than new users of ISD and HCQ, would have introduced bias. Specifically, there would have been significant differences in duration of drug exposure both between and within drug groups during which time the outcome would not have been assessed. While we had all available claims for each individual within each commercial and Medicaid datasets, if an individual was dually insured, it is possible that claims may have been missing. Despite the use of propensity-score matching, there may be residual confounding from a number of potential factors unavailable in our claims data including race/ethnicity, socioeconomic status, and SLE duration, severity and activity. These characteristics may influence who receives an ISD compared with HCQ, as well as the overall risk of cervical cancer. However, it is important to note that SLE disease severity has been shown to not be a susceptibility factor for the development of cervical intraepithelial neoplasia.19 In addition, there may be a dose-dependent effect, an interaction effect between different medications, or a greater risk conferred by one specific ISD over another. However, due to the low incidence of the outcome, we were underpowered to conduct further analyses to examine these factors. While we excluded patients who received ISD or HCQ in the 365 days prior to the index date, we were unable to assess use prior to that point and it is plausible that a remote history of ISD use may also contribute to risk. In addition, we are unable to draw conclusions about long-term cervical cancer risk associated with ISD use because of the relatively short follow-up time observed in both cohorts while patients were actively receiving the drug.
Overall, in this nationwide study of more than 20,000 diverse SLE patients, while we found a trend towards an increased rate of cervical dysplasia and cervical cancer among ISD initiators compared with HCQ, we did not find a statistically significant difference. Despite our large sample size, our power to detect clinically meaningful differences was limited, especially since the incidence of high-grade cervical dysplasia and cervical cancer was low in both of our cohorts. Further studies are needed with extended follow-up time to determine whether prolonged duration of therapy contributes to increased risk. In addition, with the known safety and efficacy of the HPV vaccine, the trend observed in this study may serve as further impetus to ensure that SLE patients, especially those receiving ISD, are appropriately vaccinated for this preventable disease.
Acknowledgments
Candace H. Feldman: Funded by the Rheumatology Research Foundation Investigator Award. She also receives research support from Pfizer Pharmaceuticals.
Daniel H. Solomon: Supported by the NIH grants K24 AR055989, P60 AR047782 and R01 AR056215. Dr. Solomon has received research support from CORRONA, Astra Zeneca, Amgen, Pfizer, Genentech, Bristol Myers Squibb, and Lilly. He also serves in an unpaid role on studies sponsored by Pfizer.
Seoyoung C. Kim: Funded by NIH grant K23 AR059677. Dr. Kim also receives research support from Pfizer, Astra Zeneca, Lilly and Genentech.
Footnotes
Disclosures:
Jun Liu: No disclosures
Sarah Feldman: No disclosures
Author Contributions: The authors listed made a substantial contribution to the concept and design, acquisition of data or analysis and interpretation of data (CHF, JL, SF, DHS, SCK), drafted the article (CHF, SCK) or revised it for important intellectual content (JL, SF, DHS) and approved the version to be published (CHF, JL, SF, DHS, SCK)
References
- 1.Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin. 2012;62:10–29. doi: 10.3322/caac.20138. [DOI] [PubMed] [Google Scholar]
- 2.Forhan SE, Gottlieb SL, Sternberg MR, et al. Prevalence of sexually transmitted infections among female adolescents aged 14 to 19 in the United States. Pediatrics. 2009;124:1505–1512. doi: 10.1542/peds.2009-0674. [DOI] [PubMed] [Google Scholar]
- 3.Mitchell MF, Tortolero-Luna G, Wright T, et al. Cervical human papillomavirus infection and intraepithelial neoplasia: a review. J Natl Cancer Inst Monogr. 1996:17–25. [PubMed] [Google Scholar]
- 4.Feldman CH, Hiraki LT, Lii H, Seeger JD, Kim SC. Human papillomavirus vaccine uptake among individuals with systemic inflammatory diseases. PLoS One. 2015;10:e0117620. doi: 10.1371/journal.pone.0117620. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kim SC, Glynn RJ, Giovannucci E, et al. Risk of high-grade cervical dysplasia and cervical cancer in women with systemic inflammatory diseases: a population-based cohort study. Ann Rheum Dis. 2015;74:1360–1367. doi: 10.1136/annrheumdis-2013-204993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Tam LS, Chan PK, Ho SC, et al. Natural history of cervical papilloma virus infection in systemic lupus erythematosus - a prospective cohort study. J Rheumatol. 2010;37:330–340. doi: 10.3899/jrheum.090644. [DOI] [PubMed] [Google Scholar]
- 7.Kim SC, Gillet VG, Feldman S, et al. Validation of claims-based algorithms for identification of high-grade cervical dysplasia and cervical cancer. Pharmacoepidemiol Drug Saf. 2013;22:1239–1244. doi: 10.1002/pds.3520. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gagne JJ, Glynn RJ, Avorn J, Levin R, Schneeweiss S. A combined comorbidity score predicted mortality in elderly patients better than existing scores. J Clin Epidemiol. 2011;64:749–759. doi: 10.1016/j.jclinepi.2010.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Mangione-Smith R, McGlynn EA, Hiatt L. Screening for chlamydia in adolescents and young women. Archives of pediatrics & adolescent medicine. 2000;154:1108–1113. doi: 10.1001/archpedi.154.11.1108. [DOI] [PubMed] [Google Scholar]
- 10.Austin PC. The performance of different propensity score methods for estimating marginal odds ratios. Stat Med. 2007;26:3078–3094. doi: 10.1002/sim.2781. [DOI] [PubMed] [Google Scholar]
- 11.Austin PC. Some methods of propensity-score matching had superior performance to others: results of an empirical investigation and Monte Carlo simulations. Biom J. 2009;51:171–184. doi: 10.1002/bimj.200810488. [DOI] [PubMed] [Google Scholar]
- 12.Dugue PA, Rebolj M, Hallas J, Garred P, Lynge E. Risk of cervical cancer in women with autoimmune diseases, in relation with their use of immunosuppressants and screening: population-based cohort study. Int J Cancer. 2015;136:E711–E719. doi: 10.1002/ijc.29209. [DOI] [PubMed] [Google Scholar]
- 13.Rojo-Contreras W, Olivas-Flores EM, Gamez-Nava JI, et al. Cervical human papillomavirus infection in Mexican women with systemic lupus erythematosus or rheumatoid arthritis. Lupus. 2012;21:365–372. doi: 10.1177/0961203311425517. [DOI] [PubMed] [Google Scholar]
- 14.Kane S, Khatibi B, Reddy D. Higher incidence of abnormal Pap smears in women with inflammatory bowel disease. Am J Gastroenterol. 2008;103:631–636. doi: 10.1111/j.1572-0241.2007.01582.x. [DOI] [PubMed] [Google Scholar]
- 15.Nyberg G, Eriksson O, Westberg NG. Increased incidence of cervical atypia in women with systemic lupus erythematosus treated with chemotherapy. Arthritis Rheum. 1981;24:648–650. doi: 10.1002/art.1780240503. [DOI] [PubMed] [Google Scholar]
- 16.Ognenovski VM, Marder W, Somers EC, et al. Increased incidence of cervical intraepithelial neoplasia in women with systemic lupus erythematosus treated with intravenous cyclophosphamide. J Rheumatol. 2004;31:1763–1767. [PubMed] [Google Scholar]
- 17.Klumb EM, Araujo ML, Jr, Jesus GR, et al. Is higher prevalence of cervical intraepithelial neoplasia in women with lupus due to immunosuppression? J Clin Rheumatol. 2010;16:153–157. doi: 10.1097/RHU.0b013e3181df5261. [DOI] [PubMed] [Google Scholar]
- 18.Bernatsky S, Ramsey-Goldman R, Gordon C, et al. Factors associated with abnormal Pap results in systemic lupus erythematosus. Rheumatology (Oxford) 2004;43:1386–1389. doi: 10.1093/rheumatology/keh331. [DOI] [PubMed] [Google Scholar]
- 19.Dhar JP, Gregoire L, Lancaster W, et al. Risk for Cervical Intraepithelial Neoplasia in Systemic Lupus Erythematosus is not Related to Disease Severity. Curr Rheumatol Rev. 2013;9:301–304. doi: 10.2174/1573397109666140103000955. [DOI] [PubMed] [Google Scholar]