Candida and dysbacteriosis: A cytologic, population-based study of 100,605 asymptomatic women concerning cervical carcinogenesis

Authors

  • Marian K. Engberts MD,

    Corresponding author
    1. Leiden Cytology and Pathology Laboratory, Leiden, the Netherlands
    2. Department of Obstetrics and Gynecology, Diakonessenhuis, Utrecht, the Netherlands
    • Leiden Cytology and Pathology Laboratory, P.O. Box 16084, 2301 GB Leiden, the Netherlands
    Search for more papers by this author
    • Fax: (011) 31 715123776

  • Banut S. M. Verbruggen MD, PhD,

    1. Leiden Cytology and Pathology Laboratory, Leiden, the Netherlands
    2. Department of Obstetrics and Gynecology, Diakonessenhuis, Utrecht, the Netherlands
    Search for more papers by this author
  • Mathilde E. Boon MD, PhD,

    1. Leiden Cytology and Pathology Laboratory, Leiden, the Netherlands
    Search for more papers by this author
  • Maarten van Haaften MD, PhD,

    1. Department of Surgical Gynecology and Oncology, University Medical Center Utrecht, Utrecht, the Netherlands
    2. Department of Obstetrics and Gynecology, Diakonessenhuis, Utrecht, the Netherlands
    Search for more papers by this author
  • A. Peter M. Heintz MD, PhD

    1. Department of Obstetrics and Gynecology, Diakonessenhuis, Utrecht, the Netherlands
    Search for more papers by this author

Abstract

BACKGROUND.

The objective of this study was to investigate whether the presence of vaginal Candida or dysbacteriosis predisposes women to an increased susceptibility for (pre)neoplasia over time.

METHODS.

A retrospective, longitudinal, cohort study was performed and was conducted in a population of 100,605 women, each of whom had 2 smears taken over a period of 12 years as part of the Dutch Cervical Screening Program. From these women, a cohort of 1439 women with Candida and a cohort of 5302 women with dysbacteriosis were selected as 2 separate study groups. The control cohort consisted of women who had completely normal cervical smears (n = 87,903 women). These groups were followed retrospectively over time. The odds ratios (OR) for squamous abnormalities in the follow-up smear for the women in these 3 cohorts were established.

RESULTS.

The dysbacteriotic cohort was significantly more likely to have low-grade squamous intraepithelial lesions (LSIL) and high-grade squamous intraepithelial lesions (HSIL+) in their follow-up smear (OR, 1.85; 95% confidence interval [95% CI], 1.28–2.67 and OR, 2.00; 95% CI, 1.31–3.05, respectively) compared with women in the control group. In contrast, the Candida cohort had no significantly increased or decreased risk of developing SIL. The equivocal diagnosis ‘atypical squamous cells of undetermined significance’ was rendered significantly more often in the follow-up smear of both study cohorts (Candida cohort: OR, 1.42; 95% CI, 1.03–1.95; dysbacteriotic cohort: OR, 1.44; 95% CI, 1.22–1.71).

CONCLUSIONS.

The results from this study indicated that the presence of Candida vaginalis was not associated with an increased risk for SIL over time. In contrast, women with dysbacteriosis had a significantly increased risk of developing (pre)neoplastic changes. These findings should be taken into account in further research concerning predisposing factors for cervical carcinogenesis. Cancer (Cancer Cytopathol) 2007. © 2007 American Cancer Society.

Vaginal complaints account for large numbers of visits to general practices in the Netherlands. Approximately 50 per 1000 women per year visit their general practitioner with complaints of vaginal discharge that is abnormal in amount, color, and/or smell.1 Possible causes of the vulvovaginal discomfort include bacterial vaginosis and vulvovaginal candidiasis (VVC).1, 2 Roughly 75% of all sexually active women will experience 1 episode of VVC during their lives, but Candida also can be isolated from the genital tract of 20% of asymptomatic women during their childbearing years.3, 4 To date, the mechanism by which Candida organisms transform from commensal to pathogen has not been established. It has been suggested that the normal vaginal flora, dominated by presence of lactobacilli, is protective against Candida vaginitis.5 However, VVC occurs more frequently in women with a lactobacilli-predominated vaginal flora and is correlated inversely with a vaginal flora change with a mixed, anaerobic vaginal flora.6 In contrast, Bacterial vaginosis is characterized by a lack of normally protective lactobacilli and an overgrowth of mainly anaerobic bacteria.7, 8 Because lactobacilli are part of the main defense mechanisms of the vagina, changes in the proportions of these Lactobacillus species are risk factors for developing vaginal infections.9–14 In addition, studies have noted that cervical cytologic abnormalities occur more often in women who have abnormal vaginal flora than in women without this condition.15–18 Therefore, we were interested in investigating whether women who carry Candida are prone to acquire cervical cytologic abnormalities over time compared with women who have a known, disturbed bacterial vaginal flora.

The Dutch national coding system for pathology findings in cervical cytology (KOPAC is an acronym for this coding system in which K indicates composition [kompositie]; O, inflammation [ontstekingsverschijnselen]; P, squamous epithelium [plaveiselepitheel]; A, other abnormalities endometrium [andere afwijkingen endometrium]; C, endocervical columnar epithelium [cylinderepitheel endocervix]) was designed especially to store cytopathologic findings, including the presence or absence of Candida and dysbacteriosis.19 Dysbacteriosis represents the microscopic diagnosis of a disturbed vaginal flora in which lactobacilli are replaced by a mixture of anaerobic and aerobic bacteria (common agents include Gardnerella vaginalis, Mobiluncus species, Bacteroides spp., Mycoplasma hominis, and Ureaplasma urealyticum) and, thus, is related closely to the clinical syndrome bacterial vaginosis.17, 20 In addition, squamous epithelial changes are coded from P2/P3 (atypical squamous cells of undetermined significance [ASCUS] or borderline) to P9 (clearly invasive squamous cell carcinoma) in the KOPAC system, which allows the recording of the occurrence of Candida and dysbacteriosis in each category.

The Leiden database contains the cytologic findings from 100,605 women who had 2 cervical smears taken over a period of 12 years as part of the national screening program for cervical carcinoma. The objective of the current study was to perform a retrospective, longitudinal, population-based cohort analysis that assessed a possible association between the presence of vaginal Candida, dysbacteriosis, and the development of cervical cancer in time.

MATERIALS AND METHODS

In the period between 1991 and 2003, the Leiden Cytology and Pathology Laboratory received almost 800,000 conventional smears. These smears originated from women who participated in the Dutch National Screening Program, because all women between the ages of 30 years and 60 years in the Netherlands are invited once every 5 years to be tested for cervical cancer. All smears were coded according to the Dutch national coding system (KOPAC) for cervical cytology.19

The O in KOPAC stands for inflammatory changes (ontsteking), and this category is divided into 9 different subgroups (O codes): koilocytosis (O1), Trichomonas vaginalis (O2), dysbacteriosis (O3) (see Fig. 1), Candida (O4), G. vaginalis (O5), no inflammatory changes (O6) (see Fig. 2), Actinomyces (O7), Chlamydia trachomatis (O8), and nonspecific changes (O9). The P in KOPAC stands for squamous epithelium (plaveiselepitheel). Within this category, a code of P1 indicates normal or benign epithelium, P2/P3 indicates ASCUS (borderline changes), P4 indicates mild dysplasia, P5 indicates moderate dysplasia, P6 indicates severe dysplasia, P7 indicates carcinoma in situ, P8 indicates microinvasive carcinoma, and P9 indicates macroinvasive squamous cell carcinoma. All diagnoses of P5 or greater generate a referral to the hospital for a colposcopic examination and, if required, a biopsy. For the purposes of the current study, we grouped smears that were coded from P5 to P9 into a single category (P5–P9). The relation of the KOPAC P codes to the Bethesda system is shown in Table 1. For each smear, there is an O code and a P code (P1, P2–3, P4, and P5–P9), which allowed us to study the prevalence of both inflammation and (pre)neoplastic changes. The database contained 748,940 smears from 432,064 women. We selected all women whose first smear, taken as part of the national screening program, showed normal squamous epithelium (Bethesda classification, within normal limits) and who had a subsequent smear obtained. In this group of 100,605 asymptomatic women, the women who had a diagnosis of Candida (n = 1439 women) formed the first study cohort, and the women who had a diagnosis of dysbacteriosis (n = 5302 women) formed the second study cohort. The control group consisted of women who had completely normal smears (n = 87,903 women) (see Table 2). All 3 cohorts were followed retrospectively over time. The mean (±standard deviation) length of follow-up between the subsequent smears was 4 years (±1.7 years). Odds ratios (OR) with 95% confidence intervals (95% CI) of the repeat smears were calculated separately for both the Candida cohort and the dysbacteriotic cohort (compared with the control cohort) by using SPSS software (version 10.0; SPSS Inc., Chicago, Ill).

Figure 1.

Dysbacteriosis in a Papanicolaou-stained cervical smear. Dysbacteriosis is diagnosed cytomorphologically by the absence of large, gram-positive rods (Lactobacillus morphotypes) and the presence of a coccoid bacterial flora.

Figure 2.

Lactobacilli (rod-shaped) in a Papanicolaou-stained cervical smear (normal healthy flora).

Table 1. Relation of the P Codes of the Dutch National Coding System for Pathology Findings in Cervical Cytology and Other Classification Systems
Codes of the KOPAC system*Description by other systemsBethesda system
  • P indicates squamous epithelium [plaveiselcellen]; WNL, within normal limits; ASCUS, atypical squamous cells of undetermined significance; LSIL, low-grade squamous intraepithelial lesion; HSIL, high-grade squamous intraepithelial lesion.

  • *

    KOPAC is an acronym for the Dutch national coding system for pathology findings in cervical cytology in which K indicates composition [kompositie]; O, inflammation [ontstekingsverschijnselen]; P, squamous epithelium [plaveiselcelepitheel]; A, other abnormalities endometrium [andere afwijkingeln endometrium]; C, endocervical columnar epithelium [cylinderepitheel endocervix]. Dutch language terms in square brackets.

P1NormalWNL
P2–P3Borderline changesASCUS
P4Mild dysplasiaLSIL
P5Moderate dysplasiaHSIL
P6Severe dysplasiaHSIL
P7Carcinoma in situHSIL
P8Microinvasive carcinomaCarcinoma
P9Squamous cell carcinomaCarcinoma
Table 2. Pathogens Detected in Cytologically Normal Cervical Smears From 100,605 Asymptomatic Women
O Code*No. (%) of womenCohorts followed over time
  • O indicates Ontsteking (inflammatory changes).

  • *

    O codes for the smears of 100,605 women who were screened for cervical carcinoma.

O1: Koilocytosis13 (0.01) 
O2: Trichomonas vaginalis334 (0.33) 
O3: Dysbacteriosis5302 (5.27)Dysbacteriotic cohort
O4: Candida1439 (1.43)Candida cohort
O5: Gardnerella vaginalis52 (0.05) 
O6: No inflammation87,903 (87.37)Control cohort
O7: Actinomyces801 (0.80) 
O8: Chlamyldia trachomatis171 (0.17) 
O9: Nonspecific changes4590 (4.56) 
Total100,605 (100) 

The KOPAC system, as described above, also contains a category to document the presence of G. vaginalis (O5). The cervical smears in this category are dominated by clue cells, and the adhering bacteria all have the same coccoid morphology.19 In other words, in contrast to dysbacteriosis, only 1 type of bacteria is encountered; however, similar to dysbacteriosis, lactobacilli are absent. However, because of the very small sample size (n = 52 women), the women who had only G. vaginalis in their cervical smears were not included in our current analysis.

RESULTS

Table 3 shows the distribution according to age of women in the Candida cohort, the dysbacteriotic cohort, and the control group. Note that, because cervical screening in the Netherlands targets women ages 30 to 60 years, all women were aged >30 years and <60 years. A chi-square indicated that there were significant differences in age distribution between the 3 cohorts (P < .001) (see Table 3).

Table 3. Distribution by Age
Age, yCandida cohort*Dysbacteriotic cohortControl cohort
No. (%)No. (%)No. (%)
  • *

    Chi-square test for the Candida cohort: observed versus expected = 17,858 (2 df; P < .001).

  • Chi-square test for the dysbacteriotic cohort: observed verses expected = 170,037 (2 df; P < .001).

30–40706 (49.1)2043 (38.5)41,405 (47.1)
40–50560 (38.9)2358 (44.5)32,386 (36.8)
50–60173 (12)901 (17)14,112 (16.1)
Total1439 (100)5302 (100)87,903 (100)

Table 4 shows the frequency of the different P-code outcomes from the second (or follow-up) smear in the Candida cohort, the dysbacteriotic cohort, and the control group. In the second smear, borderline changes (ASCUS) were established significantly more often in both study cohorts (Candida cohort: OR, 1.42; 95% CI, 1.03–1.95; dysbacteriotic cohort: OR, 1.44; 95% CI, 1.22–1.71). In addition, the dysbacteriotic cohort was significantly more likely to have low-grade squamous intraepithelial lesions (LSIL, OR, 1.85; 95% CI, 1.28–2.67) and high-grade squamous intraepithelial lesions (HSIL (including carcinoma [HSIL1], OR, 2.00; 95% CI, 1.31–3.05) in the second smear compared with the control group. In contrast, the Candida cohort had a lower risk of having LSIL (OR, 0.42; 95% CI, 0.11–1.70) but a slightly greater chance of developing HSIL+ (OR, 1.22; 95% CI, 0.45–3.28). However, these differences were not statistically significant.

Table 4. Squamous Abnormalities in the Second (or Follow-up) Smear
Second (follow-up) smearCandida cohortDysbacteriotic cohortControl cohort
No.OR95% CINo.OR95%CINo.
  • OR indicates odds ratio; 95% CI, 95% confidence interval; ASCUS, atypical squamous cells of undetermined significance; LSIL, low-grade squamous intraepithelial lesion; HSIL+, high-grade squamous intraepithelial lesion (including carcinoma).

  • *

    Statistically significant.

P1 (normal)13931.00Reference50971.00Reference85,673
P2–P3 (ASCUS)401.42*1.03–1.951491.44*1.22–1.711737
P4 (LSIL)20.420.11–1.70321.85*1.28–2.67291
P5–P9 (HSIL+)41.220.45–3.28242.00*1.31–3.05202
Total1439530287,903

DISCUSSION

In this study, we investigated whether presence of vaginal Candida or dysbacteriosis, as determined in Papanicolaou-stained cervical smears, plays a role in cervical carcinogenesis. Today, it is well known that infection with putative oncogenic human papillomaviruses (HPV) contributes to the development of cervical cancer.21–24 Although HPV infection is widely prevalent, only a few infected women will go on to develop cervical cancer, suggesting that other factors are involved in malignant progression. Cervical inflammation has been proposed as one of the cofactors in cervical carcinogenesis, because it has been established that disturbance of the vaginal flora increases the risk acquiring HPV infection.25–28 Papanicolaou and Traut29 and Mead30 previously confirmed that women with cervical carcinoma often had a dysbacteriotic flora, lacking the normally protective lactobacilli. In the current study, we also noted that women with dysbacteriosis in their smears had a significantly but slightly greater risk of developing (pre)neoplastic changes over time. This finding is supported in part by other studies that investigated women with dysbacteriosis.15, 17, 25, 31, 32 However, 2 studies showed similar frequencies of dysbacteriosis among women who had squamous intraepithelial lesions (SIL) compared with women who did not have SIL.33, 34 This discrepancy may be explained by the finding that the data in these studies were gathered simultaneously instead of over time. In addition, the presence of bacterial vaginosis was studied in women who already had cytologic abnormalities.

We established that the equivocal diagnosis ‘ASCUS’ was rendered significantly more often in the second smear for both study groups. According to the Bethesda system, ASCUS is defined as cellular abnormalities that are more marked than the abnormalities that are attributable to reactive changes but that fall short, either quantitatively or qualitatively, of a definitive diagnosis of SIL.35 However, the histologic correlate of this equivocal diagnosis is wide, ranging from a totally normal cervix mucosa to infiltrating carcinoma.36 In addition, vaginal infections like VVC can mimic the cytomorphology of ASCUS by displaying nuclear enlargement and hyperchromasia in the vaginal epithelial cells.37–39 The increased ASCUS scores noted in the current study in the cervical smears from women with Candida and dysbacteriosis may have been caused by cellular side effects of a vaginal infection that still was present after 4 years. In this context, we stress the finding that, because of the wide definition of ASCUS, it is not possible to make more concrete remarks about a possible causal relation between Candida and ASCUS. Thus, randomized controlled trials will be necessary to address this possible relation.

In contrast, we established that presence of Candida is not associated with a significantly increased risk for SIL over time. This result was confirmed by other studies, which also demonstrated no obvious associations between vaginal Candida and (pre)neoplasia.18, 40, 41 Therefore, it appears that vaginal Candida is not a cofactor in the development of cervical cancer. This is consistent with the general hypothesis that the local cervicovaginal milieu plays a role in susceptibility to HPV infection, because women who carry Candida are likely to possess a healthy Lactobacillus-predominated vaginal flora in contrast to women with dysbacteriosis.21, 22

Finally, our results indicated that here were significant differences in age distribution between the Candida cohort, the dysbacteriotic cohort, and the control group. This finding was expected, because the prevalence of dysbacteriosis gradually increases for women aged <50 years and then decreases to values observed among women aged 30 years,17 whereas the prevalence of Candida declines with increasing age.18

In this study, we demonstrated that the presence of Candida vaginalis is not associated with an increased risk for SIL over time. In contrast, women with dysbacteriosis do have a significantly increased risk of developing (pre)neoplastic changes. These findings should be taken into account in further research concerning the predisposing factors for cervical carcinogenesis.

Acknowledgements

We thank Tj. Romke Bontekoe, PhD at Oegstgeest for his excellent data analysis.

Ancillary