Variations in Progression and Regression of Precancerous Lesions of the Uterine Cervix on Cytology Testing Among Women of Different Races

Methods

Results

We identified 8589 potentially eligible patients, from which 954 (11.1%) were excluded because they had an excision procedure. An additional 1333 patients (17.5%) were excluded because race was not recorded in their medical record, and 830 patients (10.9%) were excluded owing to unsatisfactory Pap test results. Therefore, the analytic sample included 5472 women receiving a Pap test between January 2006 and September 2016, contributing a total of 24,316 person-years of follow-up time. Descriptive statistics are presented in Table 1 and Table 2. The mean (SD) baseline age was 34.5 (12.1) years (range, 13-91 years) for those included in the sample. The sample was racially diverse, with 2019 (36.9%) self-reporting as Hispanic, 1160 (21.2%) as black, 389 (7.1%) as Asian, and 1878 (34.3%) as white.

The majority of patients (3420 [62.5%]) received a diagnosis of ASC-US during follow-up; 651 (11.9%) ever had a diagnosis of ASC-H; 1882 (34.4%) ever received an LSIL diagnosis; and 290 (5.3%) ever had a diagnosis of HSIL. A total of 2374 patients (43.4%) contributed 2 or more observations to the study, meaning that the majority of patients were censored on all progressions and regressions. The distributions of changed diagnoses and person-time contributed are shown in Figure 1. Table 3 presents average times to progression or regression to and from each of the abnormal Pap test results, along with the number of patients who contributed time to each wait time estimate.

Figure 1.

Distributions of changed precancerous cervical lesion diagnoses and person-time contributed. Abbreviations: ASC-H, atypical squamous cells cannot exclude high-grade squamous intraepithelial lesion; ASC-US, atypical squamous cells of undetermined significance; HSIL, high-grade squamous intraepithelial lesion; LSIL, low-grade squamous intraepithelial lesion.

Table 4 shows results of the multistate model with race as a covariate. As stated in the Methods section, Native American (n=0), Pacific Islander (n=1), and other (including multiracial) (n=25) patients, these categories were dropped from the regression model because of the very low patient counts. We tested the global hypothesis that a model including race is more informative than an intercept-only model using a Pearson χ² likelihood ratio test for nested models. This test resulted in statistical significance (P=.004), indicating that differences in wait times by race are not entirely due to chance.

Hazard ratios and 95% CIs are presented in Table 4 and shown in Figure 2, representing the ratio of the rate of each transition for each race group compared with the rate for the same transition in the white group. Of 21 hazard ratios tested for significance, the following 4 hazard ratios (95% CIs) were statistically significant: ASC-US to LSIL for Hispanics (0.72; 95% CI, 0.54-0.96), LSIL to ASC-US for Hispanics (1.55; 95% CI, 1.04-2.31), LSIL to ASC-US for Asians (1.91; 95% CI, 1.08-3.36), and HSIL to LSIL for blacks (0.39; 95% CI, 0.16-0.96). A trend seemed to suggest that all racial groups other than whites had a longer rate of progression from ASC-US to LSIL, with Hispanics having the slowest rate, and a shorter rate of progression from LSIL to HSIL, with Hispanics having the fastest rate. With regard to regression, blacks had the slowest rate of regression from HSIL to LSIL, and Asians had the fastest rate from LSIL to ASC-US.

Figure 2.

Hazard ratios and 95% CIs for progression and regression of precancerous lesions of the uterine cervix in women of different races. Of 21 hazard ratios tested for significance, the following 4 hazard ratios were statistically significant: ASC-US to LSIL for Hispanics (0.72; 95% CI, 0.54-0.96), LSIL to ASC-US for Hispanics (1.55; 95% CI, 1.04-2.31), HSIL to LSIL for blacks (0.39; 95% CI, 0.16-0.96), and LSIL to ASC-US for Asians (1.91; 95% CI, 1.08-3.36). Abbreviations: ASC-US, atypical squamous cells of undetermined significance; HSIL, high-grade squamous intraepithelial lesion; LSIL, low-grade squamous intraepithelial lesion.

Figure 3 plots expected times to progression and regression by racial group, making it possible to interpret overall time from ASC-US to HSIL and vice versa. The Hispanic group demonstrated the fastest progression (expectation, 17.6 years; 95% CI, 11.5-25.5), as well as the fastest regression (expectation, 27.6 years; 95% CI, 21.5-35.6). The black group had the slowest expected times for both progression (expectation, 28.1 years; 95% CI, 14.6-47.2) and regression (expectation, 49 years; 95% CI, 29.1-86.2).

Figure 3.

Expected times to (A) progression and (B) regression of precancerous cervical lesion diagnoses by racial group. The Hispanic group demonstrated the fastest expected progression (17.6 years; 95% CI, 11.5-25.5), as well as the fastest regression (27.6 years; 95% CI, 21.5-35.6), and the black group has the slowest expected times for both progression (28.1 years; 95% CI, 14.6-47.2) and regression (49 years; 95% CI, 29.1-86.2). Abbreviations: ASC-US, atypical squamous cells of undetermined significance; HSIL, high-grade squamous intraepithelial lesion; LSIL, low-grade squamous intraepithelial lesion.

In Table 5, results are presented for the assessment of selection bias. The number of visits (1 vs ≥2) contributed to the study was differentially distributed both by race (P=.033) and by last diagnosis (P<.001). Patients whose last diagnosis was ASC-H were much more likely to have contributed only 1 visit. Patients whose records lacked information on race (17.9% of the collected sample) were also more likely to have contributed only 1 visit than patients with other last diagnoses (70% vs 52%-60%).

Discussion

Although racial differences in cervical cancer and related mortality have been consistently observed, no study, to our knowledge, has examined the variation in progression and regression of precancerous lesions of the uterine cervix on cervical cytology testing among patients of different races who received consistent and adequate follow-up care according to current ASCCP guidelines.

The current study found that in a 10-year sample of patients, the rates of progression and regression of precancerous lesions of the uterine cervix were not uniform across races. We also thoroughly reviewed the studies used to revise the 2006 ASCCP consensus guidelines and create the 2012 guidelines for the management of abnormal cervical cancer screening test results and cancer precursors.^22-35 These studies included comments about the demographics of the populations studied; however, they did not evaluate specific differences in race.

Invasive cervical carcinoma has been found to be nearly twice as common in blacks as in whites, with age-adjusted rates of 14.8 per 100,000 and 7.8 per 100,000, respectively.³² Studies that have examined racial differences in cervical cancer–related mortality have found that an appreciable portion of the difference can be attributed to factors related to access to health care.³⁶ However, because the patients in the current study all had access to health care and adequate follow-up, our results suggest that differences in the incidence of invasive cervical carcinoma between blacks and whites cannot be solely due to health care disparities but instead may represent potential biological factors as the root of this observed variation.

A possible biological explanation is the differences in the distribution of types and intratype variants of HPV. Yamada et al¹³ examined intratype HPV-16 sequence variation in tumor samples that were collected and analyzed in an international study of invasive cervical cancer that involved 22 countries in 5 continents. Based on E6 hybridization patterns, most of the variants from European and North American samples were phylogenetically classified as European prototype, and samples from Africa contained primarily African 1 or African 2 variants. The majority of Asian variants were observed in Southeast Asia, and almost all Hispanic variants were from Central and South America or Spain. Although this study¹³ did not report direct correlations to the behavior of precancerous lesions caused by these variants, it established a foundation for the existence of HPV variations that lead to cervical cancer and gives credence to the plausibility that HPV and resultant precancerous lesions of the uterine cervix may behave differently among various ethnicities.

In addition to the type of HPV strain that leads to cervical intraepithelial neoplasia, an important risk factor for cervical cancer is coinfection with more than 1 type of HPV: an important association has been found between the number of viral types at the site of infection and the severity of the cervical intraepithelial neoplasia.³⁷ Furthermore, Soto-De Leon et al¹⁵ demonstrated that race may be a factor in HPV coinfection: the authors examined cervical samples by Pap test and DNA PCR from 1810 women with diverse sociocultural backgrounds and showed that different ethnicities demonstrated higher multiple-strain HPV coinfection rates.

Although the exact cause is not known, the lack of uniformity in precancerous lesion behavior across different races/ethnicities in the current study may affect how these groups should be respectively screened and treated. Such differences represent an important site of intervention and should be incorporated into screening guidelines to maximize the benefit over the risks of excisional procedures in the prevention of cervical cancer once further studies solidify these differences. Given that our results demonstrate, for example, that blacks have the slowest expected times for both progression and regression of precancerous lesions, it may be acceptable to choose observation or excision for persistent precancerous lesions in cases in which the decision for observation vs excision is equivocal.

Conclusion

Our study findings indicate that the dynamics of precancerous intraepithelial lesions of the cervix are likely to be patterned by race. However, for these observations to be incorporated into clinical screening guidelines, higher-powered, prospective studies are needed to generate reliable and biologically meaningful race-specific estimates of progression and regression probabilities. More studies are required to make definitive claims about variations of these lesions among races, which ultimately will be incorporated into screening guidelines to ensure that women do not receive unnecessary excisional procedures or inadequate treatment leading to a diagnosis of invasive cervical cancer.

Author Contributions

All authors provided substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data; Dr D. Martingano and Ms Renson drafted the article or revised it critically for important intellectual content; all authors gave final approval of the version of the article to be published; and all authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.