To determine rates of human papillomavirus (HPV) infections, abnormal cervical smears, and squamous intraepithelial lesions (SIL) among women with systemic lupus erythematosus (SLE).
To determine rates of human papillomavirus (HPV) infections, abnormal cervical smears, and squamous intraepithelial lesions (SIL) among women with systemic lupus erythematosus (SLE).
We investigated 30 women with SLE, 67 with abnormal smears from colposcopy clinics, and 15 community subjects with normal smears. Polymerase chain reaction results for viral DNA and HPV-16 sequencing data were correlated to cytology and colposcopic findings.
SLE and colposcopy patients were more likely (P < 0.05) to be HPV positive (15 [54%] and 37 [67%] patients, respectively) and HPV-16 DNA positive (16 [57%] and 17 [31%] patients, respectively) than community subjects (0% HPV DNA positive and 1 [6%] HPV-16 DNA positive). SLE patients were also more likely to be HPV-16 DNA positive than colposcopy patients (P < 0.05). SLE patients with a high HPV-16 viral load more frequently had SIL (n = 6) than those with a low HPV-16 viral load (n = 1; P < 0.05). HPV and HPV-16 DNA positivity were not associated with previous or current drug therapy for SLE patients. All HPV-16 DNA sequences from 6 SLE and 5 colposcopy patients were the European-type variant. Eighteen (60%) SLE patients had a previous or current cervical abnormality. At the time of study, 5 (17%) SLE patients had an abnormal cervical smear and 8 (27%) had SIL. For those diagnosed with SLE for >10 years, the rate of SIL was 44% lower than those with SLE for <5 years (odds ratio 0.56, 95% confidence interval 0.1–3.5).
UK women with a recent SLE diagnosis had disturbingly elevated levels of HPV infections (particularly with European HPV-16 variants at a high viral load), abnormal cervical cytology, and SIL.
Systemic lupus erythematosus (SLE) is a debilitating autoimmune syndrome that is most prevalent among women and is characterized by antibodies against many tissues and multiorgan involvement. The etiology of SLE is unknown, but is believed to result from genetic, hormonal, and environmental factors (1). Management of this syndrome comprises the administration of antiinflammatory and/or immunosuppressive drugs dependent upon symptom severity. In addition to the clinical manifestations of SLE itself, female patients also have to contend with a heightened risk of developing abnormal cervical smears and squamous intraepithelial lesions (SIL) of the cervix (2–7) as well as other cancers (8). It is currently unclear whether these associations are host related, SLE syndrome specific, or, alternatively, represent a permissive effect of immunosuppression on increased host susceptibility to high-cancer-risk human papillomavirus (HPV) infections (the causative agents of SIL and cervical cancer ).
High-cancer-risk HPV type 16 and 18 infections are most commonly associated with SIL/cervical cancer in industrialized nations (10). For example, in the inner-city catchments of Lambeth and Southwark (London, UK), 68% of cervical cancers are HPV-16 DNA positive and 18% are HPV-18 DNA positive (11). Previous studies of HPV infections among patients with SLE have demonstrated that wart-virus antibodies are less frequent among patients than controls (12), suggesting an inability to produce an effective immune response to HPV, and that 11% of patients had high-risk HPV infections, which are not associated with therapy (13).
In addition to being detected in SIL and cervical cancers, HPV-16 and -18 are also often found in women with no or minor cervical abnormalities (14–16). Indeed, high-risk HPVs are cured rapidly in many women, and only infrequently do such infections result in cellular transformation (17). This may be due to viral persistence in some women and rapid clearance in others (18). Alternatively, other factors, or cofactors, may be decisive for malignant progression.
There is growing evidence that one explanation as to why some high-risk HPV infections progress to SIL/cervical cancer while others do not is that intratypic HPV variants may have markedly different pathogenic potentials. Indeed, infection with non-European rather than European HPV-16 variants confers a 2–9-fold increased risk of high-grade SIL/cervical cancer (19). Similarly, we reported (via restriction fragment length polymorphism [RFLP] assays of the E5 gene of HPV-16) that cytologically normal women were more frequently infected with RFLP variant 5, whereas those with SIL usually had variant 2 (20). Additionally, viral messenger RNA transcripts (suggestive of high transcriptional activity) and a more humanized codon usage profile (implying a greater translational potential) were detected more often in samples from patients with HPV-16 variant 2 than in those with pattern 5 variants. A subsequent study (21) confirmed these observations and demonstrated that just 1 or 2 nucleotide differences between variant 5 and variant 2 in the HPV-16 upstream regulatory region were critical for viral interactions with Brn-3a, a member of the human POU family of neurologic and cutaneous tissue-restricted transcription factors (22).
Taking these considerations into account, we performed a preliminary cross-sectional study to determine whether patients with SLE in the UK 1) are at enhanced risk of HPV infections, particularly HPV-16; 2) have high viral loads of HPV-16; 3) are infected with more pathogenic HPV-16 variants; 4) acquire HPV infections as a result of immunosuppressive drug therapy; and 5) have high levels of significant cervical disease (SIL). We investigated patients with SLE and compared them with 2 groups of participants who were at very high risk or very low risk of developing cervical premalignant disease. We anticipated that the results of this study would provide crucial statistical information for the rational design of subsequent larger cross-sectional and longitudinal studies of patients with SLE. The results demonstrated that patients with SLE are at significantly heightened risk of HPV-16 infections and of developing cervical abnormalities, particularly SIL. We were unable to demonstrate a relationship between drug therapy and HPV infection or SIL; however, this is probably a reflection of the relatively low number of patients studied.
Female patients (local and third-degree referrals from other UK regions) who had clinical signs of SLE that fulfilled the American College of Rheumatology recommendations (23) were randomly recruited from lupus clinics at St Thomas' Hospital, London, UK between 1997 and 1998 (Table 1). A comparator group of volunteers at high risk of cervical disease (all with abnormal cervical smears) was obtained from colposcopy clinics at St Thomas' Hospital. In addition, a group of community subjects were studied; all had normal cervical smears, were participating in an unrelated study, and were from the same locality as the colposcopy group. No other demographic data were available for the community subjects and their extracted cervical swab DNA samples were provided by Dr. J. M. Best (St Thomas' Hospital, London, UK). For all SLE and colposcopy patients, medical histories and a detailed questionnaire were obtained, ectocervices and endocervices were swabbed with an Axibrush (Colgate Medical, Berkshire, UK) to obtain HPV DNA for laboratory analyses, and cervical smear tests and colposcopic examinations were performed. Ethical permission for this investigation was provided by the Research Ethics Committee of St Thomas' Hospital.
|Age, median (range) years||39 (24–64)||29 (17–56)||< 0.05|
|White||27 (90)||43 (64)||< 0.05|
|African Caribbean||3 (10)||22 (33)|
|Asian||0 (0)||2 (3)|
|Single||10 (33)||43 (64)||< 0.05|
|Married||20 (67)||24 (36)|
|1–2||19 (63)||31 (46)||NS|
|≥3||3 (10)||7 (12)||NS|
|Miscarriage, >1||9 (33)||9 (13)||NS|
|Termination of pregnancy||1 (3)||13 (17)||NS|
|Number of partners|
|<5||28 (94)||41 (61)||< 0.05|
|6–10||2 (6)||20 (30)|
|>10||0 (0)||6 (9)|
|History of STDs||1 (3)||5 (7.5)||NS|
|Previous genital HPV infection||1 (3)||4 (6)||NS|
For SLE and colposcopy samples, Axibrush (Colgate Medical) tips were cut off and deposited into bijous containing 5 ml of sterile Dulbecco's phosphate buffered saline (PBS), vortexed for 1 minute, then divided into 5 × 1–ml aliquots and frozen at −20°C. Cellular debris from each 1-ml sample was pelleted by centrifugation (10,000 × g at room temperature for 10 minutes); resuspended in 200 μl of a solution containing 0.45% (volume/volume) Nonidet P40 (Sigma, Poole, UK), 0.45% (v/v) Tween 20 (Sigma), and 60 gm/liter proteinase K (Boehringer Mannheim, Lewes, UK); and then incubated overnight at 55°C. After proteinase K inactivation (10 minutes at 90°C), aliquots were stored at −70°C: 5 μl of proteinase K digest was used for each polymerase chain reaction (PCR).
To confirm samples contained DNA of sufficient quantity and quality for analyses, and to verify the absence of nonspecific PCR inhibitors, all samples were tested in β-globin PCRs. These assays utilize GH20 and PC04 primers and Thermophilus aquaticus (Taq) DNA polymerase to amplify a 286-bp segment of the human β-globin gene (24). Only samples PCR positive for β-globin DNA were tested for HPV and HPV-16 DNA.
HPV DNA was amplified using consensus MY09/MY11 HPV-degenerate primers and Taq polymerase as described previously (25). This PCR amplifies (dependent upon HPV type) ∼450 bp of the L1 gene of multiple HPV types and has an analytic sensitivity of ∼250 copies of HPV-16 DNA (26, 27).
The 2 HPV PCRs can provide 4 combinations of results. A double negative indicates the probable absence (i.e., <250 copies) of any HPV infection; HPV DNA positive/HPV-16 DNA negative indicates the presence of an HPV other than type 16 (at >250 copies); HPV DNA negative/HPV-16 DNA positive indicates HPV-16 is present at <250 DNA copies per PCR reaction (i.e., suggestive of a low HPV-16 viral load); and double-positive indicates HPV-16 at >250 genomic copies per PCR reaction (implying a high HPV-16 viral load).
To confirm the identity of the HPV-16 PCR amplicons and to establish whether particular HPV-16 intratypic variants were present in SLE patients, representative HPV-16 PCR–positive SLE and colposcopy samples as well as HPV-16 reference isolates were reamplified using the E7 primer set using Pyrococcus furiosus DNA polymerase (Promega, Madison, WI), which has 3′→5′ exonuclease activity, and amplicons were sequenced commercially (Lark UK Ltd, Takeley at Essex, UK). Resulting DNA electropherograms and sequences were inspected visually and analyzed by BLAST searches (28). Sequences were aligned with HPV-16 E7 variants downloaded from PubMed (29) and edited to the same length (271 bp) before being compared, bootstrapped, and used to construct a phylogram using ClustalW 1.82 (30).
Precautions against PCR contamination included the use of 4 geographically remote rooms (first room for preparation of PCR reagents, second for addition of clinical samples, third for amplification, and fourth for agarose-gel electrophoresis). For all PCR tests, positive controls (HPV-16–containing SiHa cells and pAT-16 DNA) and negative controls (PCR mixes devoid of target DNA and vials containing PBS exposed to clinic air) were processed and PCR-cycled in parallel with clinical samples. For individual PCR tests to be accepted, all controls had to give the appropriate result: if discrepancies occurred, the entire test was discarded and repeated using new reagents on a fresh aliquot of sample.
Thirty women with SLE (26 with case notes), 67 colposcopy patients, and DNA samples from 15 community subjects were studied. SLE patients were significantly older and more likely to be white, married, and have less than 5 sexual partners than the colposcopy patients (all P < 0.05) (Table 1).
Samples from 2 patients with SLE, 12 colposcopy patients, and no community subjects were β-globin DNA negative (Table 2). For the SLE samples, 15 (54%) were HPV DNA positive and 16 (57%) were HPV-16 DNA positive, whereas 37 (67%) colposcopy samples were HPV DNA positive and 17 (31%) were HPV-16 DNA positive. Only 1 (16%) community sample was PCR positive (for HPV-16 DNA). Compared with community samples, SLE patients were more likely to be both HPV DNA positive and HPV-16 DNA positive (i.e., have a high viral load; P < 0.01), whereas colposcopy patients were more likely to be only HPV DNA positive (P < 0.01).
|S04||Mild dys 1991||−||−||−||−|
|S30||Mild dys 1994||−||LGSIL||+||+||+|
|S09||Border 1995||−||Sq met||−||+|
|S27||Border 1996||Mild dys||−||+||+|
SLE patients with SIL were more frequently HPV-16 DNA positive than those without SIL, but this was not significant (P > 0.05). Colposcopy patients with SIL were significantly more often HPV-16 DNA positive than those without SIL (P < 0.01). Compared with the community samples, SLE patients with either any grade of SIL or high-grade SIL were more likely to have a high HPV-16 viral load (both P < 0.01). Ten (36%) SLE patients had a high HPV-16 viral load, 7 of these patients had current abnormal histology (6 had SIL), and 6 (21%) had a low HPV-16 viral load (1 with SIL; P < 0.05). Among SLE patients, being HPV or HPV-16 DNA positive was not associated with current or previous drug therapy (all P > 0.05).
HPV-16 E7 DNA sequences were obtained for 6 SLE patients (patients S06, S20, S22, S24, S25, and S30) who exhibited a wide range of current cervical disease (HPV-like histologic changes to SIL-II) and for 5 colposcopy patients (patients C05, C07, C10, C11, and C12), as were SiHa cell DNA (#37) and pAT-16 (n = 3) (Figure 1); these sequences were aligned with HPV-16 variants. The resulting phylogram demonstrated that clinical samples from SLE and colposcopy patients all had the European HPV-16 intratypic variant described by Seedorf et al (K02718) (33). The SiHa sequence (#37) matched the original description (c.f., AF001599), as did pAT-16 (c.f., K02718). Because all DNA sequences were identical to those already archived in data banks, none were submitted for accession numbers.
Overall, 18 (60%) of 30 SLE patients had a history of or current abnormal cervical smear or SIL (Table 2). Of 30 SLE patients, 5 (17%) currently had an abnormal smear and 8 (27%) had biopsy-confirmed SIL. Duration of SLE (<5 years in 11 patients and >5 years in 5 patients) was not associated with developing an abnormal smear (P > 0.05) or SIL (P > 0.05), but for patients with SLE for >10 years (n = 10), the rate of SIL was 44% lower than for those with SLE for <5 years (OR 0.56, 95% CI 0.1–3.5). All colposcopy patients had abnormal smears and 38 (56%) had SIL; the remaining patients had inflammation, HPV-like changes, squamous metaplasia, or no SIL (data not shown).
All SLE patients had received immunosuppressive therapy (85% prednisolone, 77% hydroxychloroquine, 35% azathioprine, 27% cyclophosphamide, and 12% methotrexate) (Figure 2). It should be noted that patients treated with prednisolone were all receiving very low dosages (median 0.13 mg/kg/day, range 0.02–0.3 mg/kg/day) and were unlikely to be immunosuppressed as a result. SLE patients with normal cervical smears were more likely to be currently receiving hydroxychloroquine compared with those with an abnormal smear (P < 0.01). No other associations between cervical disease and treatment were identified. Unadjusted OR for the effect of treatment on abnormal smear test results or SIL could not be determined because all patients received immunosuppressive therapy. However, when those receiving hydroxychloroquine were excluded, the OR for the effects of other therapies on SIL was 1.4 (95% CI 0.1–16). Interestingly, of 9 patients (patients S01, S05, S07, S10, S12, S16, S18, S21, and S30) who had previously received any form of drug therapy (predominantly azathioprine 75–150 mg [n = 4] or intravenous cyclophosphamide [n = 5]), 7 (78%; patients S05, S07, S10, S16, S18, S21, and S30) had evidence of current HPV infection and/or current cervical lesions. Oral contraceptive use (ever) was associated with an increased risk of abnormal histology for both SLE patients (OR 2.4, 95% CI 0.4–15.3; n = 21) and colposcopy patients (OR 2.3, 95% CI 0.7–8.1; n = 54) (data not shown).
This study demonstrated that UK women with SLE had a very high prevalence of HPV-16 European intratypic variant infections at high viral load and confirmed previous reports of increased levels of cervical disease. Patients with SLE even had a greater prevalence of HPV-16 DNA than the colposcopy patients (57% versus 31%; P < 0.05). This finding, however, may well be artifactual, reflecting ethnic differences between these 2 groups since the colposcopy group contained more African Caribbean patients. This is of pertinence, because although HPV-16 and HPV-18 are the most common causes of cervical disease in industrialized countries (10), HPV types 53 and 58 predominate in Africa (34). This conclusion is supported by the observation that more colposcopy patients had detectable HPV DNA per se (67% versus 54%) and more SIL (56% versus 27%) than SLE patients, and may reflect the fact that the latter patient group was at a much lower risk of contracting sexually transmitted diseases (SLE patients were significantly older, more likely to be married, and had a self-reported history of fewer sexual partners).
Samples from 10 patients with SLE had high HPV-16 viral loads (i.e., were positive by both PCRs) and many of them also had abnormal cervical histology (6 had SIL). Conversely, the 6 SLE patient samples that had low HPV-16 viral loads had correspondingly low levels of cervical disease (1 had SIL). Five patients were HPV positive but HPV-16 DNA negative, indicating they were infected by HPVs other than type 16 (none had SIL). Seven SLE patients whose samples were dually negative by both PCRs also had low levels of cervical disease (1 with SIL). No associations between HPV detection and treatment modality were detected, a finding reported by others (13).
HPV-16 DNA sequence analyses were performed on some SLE and colposcopy samples to determine whether particular HPV-16 variants were preferentially present in SLE patients. While the DNA sequences were relatively short (271 bp), we were able to demonstrate via alignments and phylogenetic analyses that all samples were identical to the European HPV-16 variants, which are believed to confer a lower cervical cancer risk than African-type variants. Whether this means that patients with SLE are more likely to have self-resolving premalignant cervical neoplasias than to progress to cervical cancer is unknown.
Overall, 60% of patients with SLE had a history or current evidence of cervical disease. At the time of the study, 17% of patients with SLE had an abnormal smear compared with the overall background figure of 5% for the general UK population (35). This level of abnormal cytology rate is at the lower end of published abnormal smear rates for patients with SLE, which range from 16% (13) through 24% (2, 5) to 36% (3). Conversely, our finding that 27% of patients with SLE had SIL is higher than previous studies. For example, Tam and coworkers (13) reported that 12% of patients with SLE had SIL, whereas another group found that the incidence of SIL over a 3-year period was 9.8% (7). These differences in reported SIL rates may well be explained by demographic differences between these studies: the study by Tam et al included only ethnic Chinese patients, 93% of whom were married (13). In contrast, our patients with SLE were predominantly white and only 67% were married. Similarly, patients in the study by Ognenovski et al (7) were younger than our patients (33 years versus 39 years) and the study excluded many patients, including 10 with abnormal cervical histology; if these patients had been included, their SIL rate would have been 22.5% (16 of 71), which approaches our observed rate of 27%.
In the study by Ognenovski et al (7), most patients with SLE developed SIL within 3 years after SLE diagnosis. A similar trend was identified in the present investigation in that women with SLE diagnosed for 5–10 years or for >10 years were at reduced likelihood (62.5% and 44%, respectively) of developing SIL than those diagnosed with SLE more recently (<5 years). Whereas we noted a history of cervical abnormalities (abnormal cytology and/or SIL) in 37% of our patients with SLE, another group found that 45% had a history (5). The fact that more of our patients with SLE had cervical dysplasia rather than dyskaryosis could indicate that the lesions in patients with SLE were small and were missed in the cervical smear collection, which would tie in with the detection of the less-pathogenic European HPV-16 variants.
SLE patients with normal smears were more likely to be receiving hydroxychloroquine than those with abnormal smears (P < 0.01). Furthermore, when hydroxychloroquine was excluded, the use of other medications was associated with a slightly increased OR for SIL (OR 1.4, 95% CI 0.1–16). Because hydroxychloroquine is used to manage less-aggressive SLE symptoms, these findings suggest that cervical disease is associated with either severe SLE or treatment with more potent drugs. No other associations between therapy and cervical disease were detected. This finding contrasts with another report that found that in women currently receiving azathioprine, or who had a history of receiving azathioprine, 50% had cervical atypia compared with 11% of controls (2). Other researchers have reported that women treated with cyclophosphamide were more likely to have SIL compared with those who were untreated (4). In the study by Ognenovski and colleagues (7), women with SIL were more likely to be treated with cyclophosphamide/prednisolone or cyclophosphamide/prednisolone/azathioprine than prednisolone or prednisolone/azathioprine. Indeed, the OR for increased risk for developing an abnormal cervical smear if treated with immunosuppressive therapy has been estimated to be 1.6 (95% CI 1.0–2.7) (6). Therefore overall, and in contrast to the present study, the consensus in the literature suggests that immunosuppressive treatment probably does contribute to the development of cervical disease in patients with SLE. Studies of SLE larger than the present study will be needed to unequivocally prove this point. Other autoimmune conditions are managed using similar therapeutic regimens (e.g., Crohn's disease, ulcerative colitis, and rheumatoid arthritis) and azathioprine/cyclophosphamide therapy has been associated with a higher prevalence of malignancies in rheumatoid arthritis (36).
In the present study, use (ever) of oral contraceptives was associated with cervical dysplasia among SLE and colposcopy patients (OR 2.4 and 2.3, respectively); these findings closely agree with those of Bernatsky et al for patients with SLE (OR 2.9) (6). This presumably represents an increased risk of HPV infection due to extended periods of no barrier protection during sexual intercourse.
The reason we did not detect any associations between HPV infection or cervical disease and therapy is unclear, but may be due to limitations such as this study 1) included only a relatively small number of patients with SLE (e.g., none had quiescent disease and were untreated), 2) was not case-controlled and relied upon high- and low-risk comparator groups, 3) was cross-sectional, and 4) was potentially confounded by SLE patients self-selecting as a result of prior cervical disease.
Despite these caveats, the present study provides solid evidence that women with SLE are at a considerably enhanced risk of acquiring HPV-16 infections and developing cervical premalignancies. Thus, there are strong grounds for establishing a multicenter investigation of the risk of HPV infections and cervical disease in UK women with SLE. This investigation should include analyses of all high-risk–type infections, utilize quantitative PCR methods to confirm the association between SLE and viral load, and be case-controlled with a longitudinal element. A pivotal observation of this study was that women with SLE were 3 times as likely to have an abnormal smear compared with the general UK population and were at a significantly heightened risk of developing SIL. This conclusion has a practical implication for the treatment of women with SLE: they should be investigated regularly for cervical abnormalities, particularly within the first 5 years after SLE diagnosis.
Dr. Cason had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Hughes, Shepherd, Cason.
Acquisition of data. Nath, Mant, Luxton, Hughes.
Analysis and interpretation of data. Hughes, Raju, Cason.
Manuscript preparation. Nath, Cason.
Statistical analysis. Cason.
We thank the staff of the St Thomas' SLE and colposcopy clinics, particularly Dr. M. Khamashta, Mr. A. Kubba, Sister A. Jokhan. Dr. J. Best, Ms E. Cutori, and Dr. A. Herbert, who kindly provided the community samples with negative cervical smears. We are also grateful to all study participants for their cooperation.