This HPV involvement forms the biological basis of additional patterns recently described in studies of the natural history of CIN lesions, called early regression, fluctuation, and late regression (
3,
20,
21,
23,
24). It seems obvious that the HPV type, viral load, the acquisition of new (incident) infections, as well as clearance of the virus are salient features of the natural history of cervical HPV infections (
2,
14,
21,
22,
24,
30,
31); but data on their role as risk factors in cervical carcinogenesis have only recently emerged (
1,
4). Thus, the first few studies that have addressed the mechanisms of viral clearance reported conflicting findings (
12,
16), but these were sorted out in part in our recent study (S. Syrjänen, I. P. Shabalova, N. Petrovichev, V. P. Kozachenko, T. Zakharova, A. Pajanidi, J. I. Podistov, G. Chemeris, L. G. Soazeva, E. V. Lipova, I. Tsidaeva, O. Ivanchenko, G. Pshepurko, S. Zakharenko, R. Nerovjna, L. B. Kljukina, O. A. Erokhina, M. F. Branovskaja, M. Nikitina, V. Grunberga, A. Grunberg, A. Juschenko, P. Tosi, M. Cintorino, P. Santopietro, and K. J. Syrjänen, submitted for publication). Similarly, data on the accumulation of incident HPV infections and Pap smear abnormalities are still fragmentary, and the factors predicting these events remain partly controversial (
5,
6,
9,
10,
19).
RESULTS
Table
1 depicts the acquisition events (abnormal Pap smear result, infection HR HPV, both, or neither) related to the status of the women at the baseline. The probability of acquiring an abnormal Pap smear result was significantly (
P = 0.0001) associated with the presence of HR HPV DNA at the baseline: 111 of 184 (60.3%) of the women in group 2 and 22 of 118 (18.6%) of the women in group 1 (OR, 6.63; 95% CI, 3.83 to 11.49). On the other hand, an abnormal Pap smear result at the baseline did not increase the rates of acquisition of incident HR HPV, which were identical for the women in group 1 and the Pap women in group 3: 10.2 and 9.1%, respectively (
P = 0.475).
The times of acquisition of an incident abnormal Pap smear results for women in group 1 and group 2 were statistically significantly different: 19.4 months (range, 0.83 to 34.07) and 9.2 months (range 0.9 to 34.5 months), respectively (P = 0.00001, Mann-Whitney U test). This includes the 12 women for whom an incident abnormal Pap smear result was associated with an incident HR HPV infection. The times of acquisition of incident HR HPV infection in groups 1 and 2 were 16.6 months (range, 11.5 to 31.0 months) and 11.0 months (0.8 to 34.6 months), respectively (P = 0.006). Thus, in the women in group 1, it took some 3 months longer to acquire an abnormal Pap smear result than an incident HR HPV infection (19.4 and 16.6 months, respectively) (P = 0.187). On the other hand, the rate of acquisition of an incident abnormal Pap smear result in women in group 2 was more rapid than the rate of acquisition of an incident HR HPV infection in women in group 3 (9.2 and 11.0 months, respectively) (P = 0.355). For the whole series, it took longer to acquire incident HR HPV infection than an incident abnormal Pap smear result (14.8 and 11.6 months, respectively) (P = 0.009) because of the rapid accumulation of Pap smear abnormalities in group 2 (HPV positive/Pap smear negative).
Table
2 shows the monthly rates of acquisition of abnormal Pap smear results and incident HR HPV infections in the three groups. The monthly rates of acquisition of incident HR HPV infection were very similar for group 1 and group 3, whereas the rate of acquisition of an abnormal Pap smear result was more than twice as high in group 2 (
P = 0.0001, Mann-Whitney U test) as in group 1. There was no difference in the monthly rates of accumulation of abnormal Pap smear results and HR HPV infection among the women in group 1 (
P = 0.186), whereas the difference between group 2 and group 3 was highly significant (
P = 0.0001) due to the high rate of accumulation of incident Pap smear abnormalities in group 2.
In life-table analyses, the cumulative incidence of abnormal Pap smear results was significantly higher in women in group 2 than in women in group 1 (Fig.
1) (
P = 0.0014), whereas no difference in the rates of accumulation of incident HR HPV infections could be observed between group 1 and group 3 (Fig.
2). Compared with the accumulation of incident HPV infection or Pap smear abnormalities as optional events (which reached rates of 30 to 40% at 24 months), the accumulation of a coexistent abnormal Pap smear result and incident HR HPV infection by the same women was significantly slower, reaching only 10% at 24 months of follow-up (life-table not shown).
We next compared the three groups of women with different patterns of acquisition (HPV infection, abnormal Pap smear result, or both) to the women whose baseline status did not change (i.e., no incident HR HPV infection or abnormal Pap smear result) during the follow-up by performing a univariate analysis with all the data collected on the questionnaire as well as the pertinent Pap smear and HPV DNA detection data. Table
3 summarizes the factors that proved to be significant in distinguishing the women with these four outcomes. The most significant factors were age, time of follow-up, patient category, and HR HPV load.
To demonstrate the age dependency of incident HR HPV infection and an abnormal Pap smear result, we performed univariate (Kaplan-Meier) survival analysis for both events. While no age dependency could be demonstrated for incident Pap smear abnormalities (
P = 0.1811) (life table not shown), the acquisition of incident HR HPV infections was significantly (
P = 0.0001) age dependent (Fig.
3).
The significant predictors by univariate analysis (Table
3) were entered into the multivariate survival (Cox) analysis by using the backward likelihood ratio approach. The only independent predictor of the incident abnormal Pap smear result (in both HPV DNA-negative and Pap-smear negative women and HPV DNA-positive and Pap smear-negative women) was the high viral load. The discriminative power was highest with a cutoff level of 20 RLUs/CO (OR, 2.050; 95% CI, 1.343 to 3.129;
P = 0.001) but lost its significance when the cutoff was raised or lowered. Viral load was not confounded by age (as a continuous variable) but maintained its independent predictive power.
Of the factors predicting incident HR HPV infection (in HPV-negative and Pap smear-negative women and HPV-negative and Pap smear-positive women), the patient category and ever having been pregnant were the only independent predictors, with P values of 0.0001 and 0.013, respectively. The crude rate ratio (RR) for the acquisition of incident HR HPV infection for patients who had had sexually transmitted diseases (STDs) was 10.177 (95% CI, 3.322 to 31.178) and remained significant even after adjustment for age (RR, 3.927; 95% CI, 1.193 to 12.921). The age-adjusted negative predictive power of ever having been pregnant was even more significant (P = 0.005; RR, 0.255; 95% CI, 0.098 to 0.664), suggesting that previous pregnancy is a significant protective factor for the acquisition of incident HR HPV infection.
DISCUSSION
The data on the risk factors that predispose women to the acquisition of incident infections and the factors that favor the clearance of prevalent HPV infections are emerging only gradually (
5,
6,
9,
10,
19,
27,
28). In our study, the cumulative incidence of HR HPV infections was not significantly increased by an abnormal Pap smear result at the baseline, with only 9.1% of women with an abnormal Pap smear result at the baseline developing incident HR HPV infection. The value is practically identical to that (10.2%) for women who were Pap smear negative at the baseline and is consonant with some recent data (
15). This is another indication that not all Pap smear abnormalities are related to HPV infection or that the abnormalities might be linked to the low-risk HPV types for which analyses were not performed in this study. Another feasible explanation might be that in these HR HPV-negative and Pap smear-positive women, HPV was already cleared at the time of the clinical visit and, thus, was unlikely to reappear during the follow-up. It is also well established that the panel of the HCII test does not contain all known HR HPV types, providing a minor possibility that some of these women might truly have been HPV positive at the baseline. Due to the wide coverage of the most prevalent HPV types by the HCII test, however, this possibility seems negligible and of no practical significance to the results of the study.
On the other hand, the accumulation of incident Pap smear abnormalities was significantly associated with the detection of HR HPV DNA at the baseline (Table
1). This confirms the observations previously reported by Liaw et al. (
9), Rozendaal et al. (
15), Moscicki et al. (
10,
11), Kjaer et al. (
8), and Schlecht et al. (
17), all of whom could link HR HPV types with the development of incident low-grade squamous intraepithelial lesion abnormalities in the Pap smears. These data strongly suggest that the detection of HR HPV DNA at the baseline is a significant risk factor for the acquisition of Pap smear abnormalities during a relatively short follow-up, thus predisposing these woman to the development of high-grade CIN and cervical cancer (
5,
7,
22,
23,
30,
31).
The factors that predict incident HPV infections have been addressed in some recent studies, but those that predict incident Pap smear abnormalities have been addressed in only a few studies (
2,
6,
10,
11,
19,
27). Sellors et al. (
19) and Deacon et al. (
2) reported that the median numbers of sexual partners in the past year and over a lifetime are risk factors for incident HPV infections. In another study, smoking, oral contraceptive use, and a new male sexual partner were predictive of incident HPV infection (
27). Moscicki et al. (
11) showed that HPV infection of less than 3 years' duration and daily cigarette smoking are risk factors for incident low-grade squamous epithelial lesion in a multivariable model. Recently, Schlecht et al. (
18) established the clear-cut association between the HR HPV DNA copy number and the development of incident squamous intraepithelial lesion in the Pap smear.
In the present analysis, the acquisition of an abnormal Pap smear result was not significantly age related, but women younger than 35 years of age accumulated Pap smear abnormalities at the same frequency as women older than 35 years of age (
P = 0.1811). This is in sharp contrast to the acquisition of incident HR HPV infection, which is significantly age dependent. By univariate survival analysis, women younger than 35 years of age accumulated incident HPV infections significantly (
P = 0.0001) more frequently than their older counterparts (Fig.
3). The acquisition rates were highest (3.7% per month) among women younger than age 20 years, which is in agreement with the recent data of Sellors et al. (
19). This indicates that the rate of acquisition of incident HR HPV infection is different from the rate of clearance of these viruses, because we (Syrjänen et al., submitted) and others (
5) have recently shown that the clearance of HR HPV is not age dependent but occurs at relatively constant rate across all age groups (Syrjänen et al., submitted). Young women acquire HPV infections at a significantly higher rate than older women; and because the viral clearance rate seems to be practically constant across the age groups (
5; Syrjänen et al., submitted), the balance between the incident cases and the cleared cases shifts in favor of the latter in certain age groups (Syrjänen et al., submitted), explaining the progressively declining prevalence rates in older women (
7,
22,
23,
30,
31). Furthermore, these different age dependencies of incident HR HPV infection and an abnormal Pap smear result are also consistent with the view that a substantial number of Pap smear abnormalities are not related to HPV infections, indicating that testing of women by the Pap smear and for HPV DNA does not necessarily screen for the same pathological condition (
25).
Another significant factor distinguishing these four groups (Table
3) is the category in which they were included in the study. An unchanged status is a typical pattern for women examined as a part of routine screening, whereas acquisition of both HPV infection and abnormal Pap smear results (or either) is characteristic of patients with STDs. Previous pregnancies were typical among those whose Pap smear or HPV infection status had not changed, as were previous abortions (59.1%). Importantly, patient category and ever having been pregnant remained independent predictors of incident HR HPV infections by multivariate analysis as well. Among the patients in this cohort, the presence of an STD can be regarded as a surrogate marker for promiscuous sexual behavior (
26), although we could not relate the incident HR HPV infection and abnormal Pap smear results directly to the number of recent, current, or casual sexual partners, as some other studies have done (
2,
19,
27). Similarly, a history of pregnancy is another surrogate marker for sexual behavior and reflects a stable or a nonstable partnership (
26). Never having had a Pap smear was typical of women who acquired both HPV infections and abnormal Pap smear results, or either of these, compared with the group whose status had not changed (Table
3).
There was a highly significant difference in the viral loads between those women who acquired incident abnormal Pap smear results and those who did not (Table
3). The viral load proved to be the only independent predictor of incident Pap smear abnormalities by multivariate analysis. The discriminative power was highest with a cutoff level 20 RLUs/CO but lost its significance when the cutoff was raised or lowered. This is in agreement with a recent report of Schlecht et al. (
18), who confirmed that the viral burden appears to have an independent effect on incident SIL. Together with the data from the univariate analysis, this suggests that both the presence of HR HPV DNA at the baseline and a high viral load are strong risk factors for incident abnormal Pap smear results during monitoring. Thus, assessment of the viral load might be helpful in identifying women at risk for high-grade CIN.
Among women who were HPV DNA negative and Pap smear negative at the baseline, the cumulative probability of incident HR HPV infection at 24 months is close to 30% and that of incident Pap smear abnormalities reaches 40% (Fig.
1 and
2). These values are comparable to those reported in some recent studies. Giuliano et al. (
6) calculated a cumulative probability of incident HR HPV infection of 32% during a 12-month follow-up, while Winer et al. (
27) reported a cumulative incidence of 32.3% at 24 months; these values are almost identical to those that we obtained in the present study. When several HR types are considered, all values for risk based on testing by HCII were clearly higher compared with the incidence of a specific type, e.g., prototype HPV type 16, which was 12.6% in 5 years (
29). Our values are also surprisingly similar to those for a cohort of 1,075 HPV-negative and Pap smear-negative women monitored by Woodman et al. (
28), who calculated a cumulative risk of 44% for any HPV infection at 3 years (Fig.
2).
For the whole cohort, it took longer to acquire HR HPV infection than an abnormal Pap smear result (14.8 and 11.6 months, respectively). This is evident from the life tables (Fig.
1 and
2), in which the survival plots for women with Pap smear abnormalities are higher than those for women with HR HPV infection. This difference, however, is due to the extremely rapid accumulation of incident Pap smear abnormalities in group 2. This higher rate of incident Pap smear abnormalities compared with the rate of HR HPV infection is another indication that a nonnegligible proportion of Pap smear abnormalities are due to causes unrelated to HPV infection. Importantly, however, in women who were HPV negative and Pap smear negative at the baseline (group 1), the time of acquisition HR HPV infection is shorter than that of an abnormal Pap smear result (16.6 and 19.4 months, respectively), indicating that HR HPV infection precedes an abnormal Pap smear result by approximately 3 months. This is consonant with our recent data on the temporal relationships of HR HPV clearance and normal Pap smear results, suggesting that the disappearance of HR HPV precedes that of an abnormal Pap smear result by 1 to 2 months (Syrjänen et al., submitted). Thus, both events are temporally closely related, but a short lag period is needed for the viral event (entry or clearance) to become manifest at the tissue level (by Pap smear or biopsy).
The times of acquisition and the monthly incidence rates of HR HPV infections are not yet generally available. Franco et al. (
5) calculated a monthly incidence rate of 1.3% for new infections, resulting in 38% cumulative HPV positivity after 18 months. This is very close to the monthly incidence rates for HR HPV infection in the present study: 1.05 and 0.81% in group 1 and group 3, respectively. In another study, the incidence rate of SIL by Pap smear was 0.73 per 1,000 women-months among HPV-negative women, but it was 8.68 per 1,000 women-months among women with HPV type 16 or 18 infections that persisted over two visits (
17). These values coincide with those from the present study, with a significantly higher monthly incidence rate of abnormal Pap smears in HPV-positive women at the baseline (Table
2).
Unlike incident Pap smear abnormalities, acquisition of HR HPV is clearly age dependent, being most frequent in the youngest women. The accumulation of incident abnormal Pap smears is significantly associated with the presence of HR HPV DNA at the baseline, and HPV DNA-positive women have a higher cumulative probability of an abnormal Pap smear result than HPV DNA-negative women, with this probability progressing at a higher monthly rate. By multivariate survival analysis, the only significant independent predictor of the incident abnormal Pap smear result was a high viral load, whereas independent predictors of incident HR HPV infection were patient category (patients with STDs) and ever having been pregnant (as a surrogate marker of sexual habits). At the baseline, incident HR HPV infection precedes an abnormal Pap smear result by approximately 3 months among HPV DNA- and Pap smear-negative women.