Natural Acquired Immunity Against Subsequent Genital Human Papillomavirus Infection: A Systematic Review and Meta-analysis

Abstract

Background. Studies have been mixed on whether naturally acquired human papillomavirus (HPV) antibodies may protect against subsequent HPV infection. We performed a systematic review and meta-analysis to assess whether naturally acquired HPV antibodies protect against subsequent genital HPV infection (ie, natural immunity).

Methods. We searched the MEDLINE and EMBASE databases for studies examining natural HPV immunity against subsequent genital type-specific HPV infection in female and male subjects. We used random-effects models to derive pooled relative risk (RR) estimates for each HPV type.

Results. We identified 14 eligible studies that included >24 000 individuals from 18 countries that examined HPV natural immunity. We observed significant protection against subsequent infection in female subjects with HPV-16 (pooled RR, 0.65; 95% confidence interval, .50–.80) and HPV-18 (0.70; .43–.98) but not in male subjects (HPV-16: 1.22; .67–1.77 [P = .05 (test for heterogeneity)]; HPV-18: 1.50; .46–2.55; [P = .15]). We also observed type-specific protection against subsequent infection for a combined measure of HPV-6/11/31/33/35/45/52/58 in female subjects (pooled RR, 0.75; 95% confidence interval, .57–.92). Natural immunity was also evident in female subjects when analyses were restricted to studies that used neutralizing assays, used HPV persistence as an outcome, or reported adjusted analyses (each P < .05).

Conclusions. HPV antibodies acquired through natural infection provide modest protection against subsequent cervical HPV infection in female subjects.

The majority of sexually active individuals acquire ≥1 alpha-genus human papillomavirus (HPV) genotype at their genital regions during their lifetime [1]. Although in most individuals the HPV infection can be cleared or controlled within 1 or 2 years [2, 3], HPV persists in the epithelium of a subset of infected individuals and can progress to cancer at several anatomic sites, particularly at the cervix [4]. At least some of the women whose HPV infection is cleared seem to lack lifelong immunity, because type-specific HPV infections can reappear among previously exposed individuals [5]. These HPV reappearances can occasionally progress at least to precancer [5, 6].

In approximately 60%–70% of women who acquire an HPV infection, a measurable type-specific serum antibody response against epitopes develops on the HPV L1 capsid protein [7, 8], representing an insensitive marker of cumulative HPV exposure [9]. Theoretically, these antibodies could help protect individuals from subsequent infections with that HPV type. Indeed, some have hypothesized that this “natural HPV immunity” may help partially explain the declining cervical HPV prevalence by age seen in some countries [10, 11] and that this immunity could be used as a factor in cost-effectiveness models evaluating HPV vaccination of older populations [12].

Protection through naturally acquired polyclonal antibodies is plausible, given that the strong antibodies responses to prophylactic HPV vaccination are believed to be responsible for the protection observed among vaccinated individuals [13]. However, the antibody levels in vaccinated women are orders of magnitude higher than natural antibody levels. Previous study findings seem to be mixed on whether individuals with HPV antibodies acquired from natural HPV infection are protected against subsequent genital HPV infection [6, 11, 14–25]. These studies were often relatively small and used different populations, assays, and analytic techniques that may have affected the discordant results. Therefore we performed a systematic review and meta-analysis of this literature to determine more definitively whether naturally acquired HPV antibodies are protective against subsequent genital HPV infection in unvaccinated individuals.

METHODS

Search Strategy and Selection Criteria

We systematically reviewed the biomedical literature for studies conducted between 1 January 1950 and 6 July 2015 using the MEDLINE and EMBASE databases. Key search terms included the following: “human papillomavirus” or “HPV” and “serology” or “viruslike particles” or “seropositivity.” Two of the authors (D. C. B. and G. J.) independently identified eligible publications by reviewing titles and abstracts, and also by searching the reference lists of eligible publications. When an abstract's content was considered inadequate to determine inclusion, the full text of the article was reviewed.

Studies included in the systematic review were required to be longitudinal and needed to evaluate “natural HPV immunity” by testing for HPV L1 serology. Studies must have compared the risk of subsequent genital HPV infection (DNA) among individuals who were HPV seropositive or HPV seronegative at the start of follow-up. We defined HPV natural immunity as HPV antibodies significantly protecting against a subsequent type-specific infection. We did not evaluate whether HPV antibodies protect against subsequent HPV-related types in the same clade (ie, HPV-16 antibodies against HPV-31/33/35/52/58 infection). We excluded any studies that did not restrict to individuals who were HPV DNA negative (for the homologous type of interest) at the entry of the analysis. We also restricted analysis to studies that examined genital HPV (DNA) infection as the outcome, because only 2 identifiable studies examined nongenital HPV infection as an outcome [20, 26]. Every other study examining natural immunity against genital HPV infection was included regardless of their study population, including those with human immunodeficiency virus (HIV)–infected individuals.

Data Extraction

We extracted data from the included studies using a standardized data-collection form. We included analyses for any alpha-mucosal HPV genotype. Every included study examined HPV DNA (infection) status measured through polymerase chain reaction (PCR)–based methods with varying primers. Many studies reported multiple outcome results; for this meta-analysis, we used the longest detected infection-level measure (up to 12-month persistence) for our main pooled analysis. For example, we used a 6-month persistent HPV infection instead of a one-time detection of HPV for the main pooled analysis when both were presented in an article. However, we also included both results in separate subanalyses specific for 6-month persistent infection and one-time detection.

Seven separate publications used the same 3 longitudinal cohort populations. Our main pooled analysis included only 1 publication from each of the 3 populations, selecting the publication using the more sensitive serologic assay (highest percentage of HPV-seropositive individuals) and the largest sample size, because we were most interested in determining the estimate of protection among all individuals with evidence of previous HPV exposure. Although no current serologic assay seems to be perfectly sensitive for HPV exposure, assays that measure a polyclonal response, such as the viruslike particle enzyme-linked immunosorbent assay (VLP-ELISA), seem more sensitive than those that only measure epitope-specific neutralizing antibodies, such as the competitive Luminex Immunoassay (cLIA) [9, 24, 27]. However, in subanalyses examining assay-specific results we included estimates from 2 other publications [15, 24] that compared results using multiple serologic assays. Most studies defined their seropositivity thresholds as 3 or 5 standard deviations above the mean titer/optical density/intensity level in concurrently tested specimens from virgins, after excluding positive outliers. Some studies presented multiple results using different seropositive cutoff estimates. We used the main cutoff in each article that was presented in the abstract and/or in tables.

Data Analysis

Random effect meta-analysis techniques were used to generate pooled summary estimates for HPV-16 and HPV-18 [28]. These 2 HPV types are known to cause approximately 70% of cervical cancer and >80% of HPV-associated noncervical cancers [2, 29, 30]. Natural immunity against 8 alpha HPV types (types 6, 11, 31, 33, 35, 45, 52, and 58) was also each analyzed in a type-specific manner, but these results were then pooled, given that the analyses were restricted to 1–4 publications, depending on the type. A global pooled estimate of natural immunity against all tested HPV types was also calculated. We used adjusted analyses in the meta-analyses when they were available, and we conducted subgroup analysis to examine whether sex may be a source of heterogeneity. We also used the longest detectable HPV infection end point when available (12-month persistent infection in 1 study and 6-month persistent infection in 3).

In addition, we performed 4 subgroup analyses, comparing (1) studies that used neutralizing serologic assays (eg, cLIA) with those that used neutralizing and nonneutralizing serologic assay (eg, VLP-ELISA); (2) studies that used MY09-MY11 consensus primer PCR with those that used the SPF10-DEIA PCR method for DNA detection; (3) studies that used a 6-month persistent HPV DNA end point with those that used one-time detection of HPV DNA as an end point; and (4) studies that reported adjusted analyses with those that reported only unadjusted analyses. We examined heterogeneity using the I² and χ² statistics [31], and publication bias using funnel plots and Egger and Begg P values [32, 33], and we performed meta-analyses with Stata SE software (version 13.1; StataCorp) [34].

RESULTS

We systematically searched 1531 articles, and excluded 1490 because the study was not longitudinal, did not test for HPV DNA, or did not test for HPV serology. The full texts of each of the remaining 41 studies were reviewed, and 14 articles were identified to meet the inclusion criteria (Figure 1).

Figure 1.

Study selection flow diagram for this systematic review and meta-analysis. Abbreviation: HPV, human papillomavirus.

The included studies spanned 18 countries and included >24 000 HPV unvaccinated individuals (Table 1). Most studies were restricted to female subjects, examining immunity against cervical HPV (11 of 14 publications, approximately 90% of total population) and the majority were conducted in the United States or Costa Rica (10 of 14 publications, approximately 55% of total population). More than half of the studies used the VLP-ELISA as their serologic assay and used one-time detection of genital HPV DNA as their outcome of interest (Table 1). Half presented an effect measure adjusted for other covariates, and the length of follow-up was ≤4 years for all but 1 study. The HPV-16 seroprevalence in the studies with female subjects ranged from 6.2% to 45.5%.

Table 1.

Description of Studies That Evaluated HPV Natural Immunity in HPV Unvaccinated Individuals

Authors (Year)	Cohort	Age Range at Entry, y	Sex	Location	Sample Size, No. of Subjects	Length of Follow-up, mo	Serologic Assay	DNA Assay Primers	DNA Outcome Measure	Effect Measure	Variables Adjusted for	HPV Types Tested	HPV-16 Seroprevalence, %	Other Notes
Castellsague et al (2014) [6]	Control arm of PATRCIA	15–25	Female	Europe, North and South America, Asia, Australia	8193	48	VLP-ELISA	SPF10-DEIA/LIPA	12-mo persistent infection	Adjusted RR	Age at first intercourse, country, marital status, lifetime number of sex partners, smoking, condom usage	HPV-16/18	15.0	Analyzed results by seropositive antibody levels: evidence of stronger protection at higher antibody level
Viscidi et al (2004) [16]	Guanacaste NHS	18–94	Female	Costa Rica	6852	84	VLP-ELISA	MY09-MY11	1-time detection	Unadjusted RR	…	HPV-16/18/31	21.8	Initial NHS study—Wentzensen et al (2011) [16] —was a follow-up
Safaeian et al (2010) [11]	Control arm of CVT	18–25	Female	Costa Rica	2813	48	VLP-ELISA	SPF10-DEIA/LIPA	1-time detection	Unadjusted RR	…	HPV-16/18	24.8	Analyzed results by seropositive level: evidence of stronger protection at higher antibody level
Velicer et al (2009) [18]	Merck Vaccine Trial	24–45	Female	South and North America, Europe, Asia	1858	30	cLIA	multiplex PCR	6-mo persistent infection	Unadjusted RR	…	HPV-6/11/16/18	18.3	HPV- 16 and HPV-18 results were presented combined, suggested potential effect modification by age
Lu et al (2012) [21]	HIM	18–70	Male	Brazil, Mexico, US	1834	18	VLP -ELISA	PGMY09/11	6-mo persistent infection	Adjusted HR	Education, alcohol, smoking, circumcision, lifetime sex partners, partner's sex	HPV-16	12.9	Mostly heterosexual men, included 211 MSM
Wentzensen et al (2011) [15]	Guanacaste NHS	18–70	Female	Costa Rica	912	84	VLP-ELISA and cLIA	MY09-MY11	1-time detection	Adjusted OR	Age, sampling adjusted	HPV-6/11/16/18	18.7 (ELISA); 9.7 (cLIA)	Suggested cLIA (neutralizing antibody) detects antibodies that are more strongly linked to immunity
Wilson et al (2014) [14]	ALTS	≥18	Female	US	886	24	Luminex-based multiplex	PGMY09/11	6-mo persistent infection	Adjusted OR	New sex partners during course of study	HPV-16/18/31/33/35/45/52/58	23.5	Suggested potential effect modification by recent sexual behavior
Mooij (2014) [20]	Netherlands MSM	≥18	Male	Nether-lands	719	12	Luminex-based multiplex	SPF10 DEIA/LIPA 25	1-time detection	Adjusted HR	HIV, age, smoking, lifetime sex partner, recent anal sex partner, anal sex position in last 6 mo, circumcision status	HPV-16/18	46.5	HIV-positive and HIV-negative MSM; genital and anal HPV-16 outcomes reported
Malik et al (2009) [19]	Rutgers University	Mean, 20	Female	US	508	36	VLP-ELISA	MY09/MY11	1-time detection	Unadjusted RR	…	HPV16/18/31/33/35/45/52/53	15.0	Results for several related alpha groups (not just specific type)
Robbins et al (2014) [23]	CVT	18–25	Female	Costa Rica	488	48	GST-L1	LiPA; SPF 10-DEIA	1-time detection	Unadjusted OR	Lifetime number of sex partners at entry	HPV-16/18	NA	CVT follow-up study from Safaeian (2010) [11] with other serologic assay
Viscidi et al (2005) [17]	HERS	26–55	Female	US	413	36	VLP-ELISA	MY09-MY11 PCR	1-time detection	Unadjusted HR	…	HPV16/18/31/35/45	45.5	Restricted to HIV-negative individuals in this analysis; this study also included HIV-positive results
Lin et al (2013) [24]	CVT	18–25	Female	Costa Rica	388	48	ELISA, cLIA, SEAP-NA	SPF 10-HPV LiPA25	1-time detection	Unadjusted OR	…	HPV-16	NA	CVT follow-up study from Safaeian (2010) [11] with other serologic assays
Lu et (2010) [22]	Tucson, AZ, men	18–44	Male	US	285	18	VLP-ELISA	PGMY09/11	1-time detection	Unadjusted IRR	…	HPV-16/18	14.4	Men, genital HPV-16/18
Ho et al (2002) [25]	Rutgers University	Mean, 20	Female	US	242	36	VLP-ELISA	PCR and Southern blot	1-time detection	Adjusted RR	Age, ethnicity, number of sex partners, duration of sexual relationship, oral contraceptives	HPV-16	6.2	High levels of IgG for HPV-16 in ≥ 2 previous visits; used HPV-related types (types 16, 31, 33, 35, 52, or 58)

Abbreviations: ALTS, ASCUS/LSIL Triage Study; cLIA, competitive Luminex Immunoassay; CVT, Costa Rica Vaccine Trial; ELISA, enzyme-linked immunosorbent assay; HERS, HIV Epidemiology Research Study; HIM, Human Papillomavirus Infection in Men Study; HIV, human immunodeficiency virus; HPV, human papillomavirus; HR, hazard ratio; IgG, immunoglobulin G; IRR, incidence rate ratio; MSM, men who have sex with men; NHS, Natural History Study; OR, odds ratio; PCR, polymerase chain reaction; RR, relative risk; SEAP-NA, secreted alkaline phosphatase neutralization assay; US, United States; VLP, viruslike particle.

In random-effect meta-analysis models including female and male subjects, we observed evidence of significant protection against subsequent genital HPV-16 DNA infection comparing HPV-16–seropositive with HPV-16–seronegative individuals (pooled relative risk [RR], 0.69; 95% confidence interval [CI], .53–.83; Figure 2). Likewise, we observed similar estimates of natural immunity for genital HPV-18 comparing HPV-18–seropositive with HPV-18–seronegative individuals (pooled RR, 0.75; 95% CI, .48–1.01; Figure 3) and for non–HPV-16/18 types (6/11/31/33/35/45/52/58) against their respective types (0.75; .57–.92; Figure 4). Combining all measured serotypes, we observed significant natural HPV immunity against their respective type-specific infection (pooled RR, 0.72; 95% CI, .62–.82).

Figure 2.

Forest plot of the relative risk for human papillomavirus (HPV) type 16 infection comparing HPV-16–seropositive with HPV-16–seronegative individuals. Overall test for heterogeneity comparing studies in male and studies in female subjects: P = .05. Abbreviations: CI, confidence interval; ES, effect size.

Figure 3.

Forest plot of the relative risk for human papillomavirus (HPV) type 18 infection comparing HPV-18–seropositive with HPV-18–seronegative individuals. Overall test for heterogeneity comparing studies in male and studies in female subjects: P = .15. Abbreviations: CI, confidence interval; ES, effect size.

Figure 4.

Forest plot of the relative risks for non–human papillomavirus (HPV) type 16 or18 infections comparing those seropositive with those seronegative to the type-specific HPV infection of interest. Abbreviations: CI, confidence interval; ES, effect size.

We observed differing results by sex. We observed significant HPV-16 and HPV-18 natural immunity in female subjects (HPV-16: pooled RR, 0.65; 95% CI, .50–.80; HPV-18: 0.70; .43–.98) but not in male subjects (HPV-16: 1.22; .67–1.77; HPV-18: 1.50; .46–2.55; Figures 2 and 3). The overall test for heterogeneity comparing female with male subjects approached significance for both HPV-16 (P = .05) and HPV-18 (P = .15).

Although there was evidence of potential heterogeneity by sex, there was no evidence for substantial heterogeneity between the 14 individual studies (HPV-16: I² = 21.7%; HPV-18: I² = 26.8%). There was also no evidence for publication bias for either the HPV-16 analyses (Egger P = .95; Begg P = .79; Supplementary Figure 1) or the HPV-18 analyses (Egger P = .84; Begg P > .99; Supplementary Figure 1).

Among studies with female subjects, in the 3 studies using assays that measured only neutralizing antibodies (eg, cLIA), the evidence of HPV-16 and HPV-18 natural immunity was nonsignificantly stronger (HPV-16: RR, 0.46; 95% CI, .24–.68; HPV-18: 0.64; .01–1.27; Figure 5) than in the studies that measured neutralizing and nonneutralizing antibodies (HPV-16: 0.64; .47–.82; HPV-18: 0.71; .39–1.04). In addition, among studies with female subjects, those that tested for HPV DNA using MY09-MY11 consensus primer PCR suggested nonsignificantly stronger natural immunity (HPV-16: RR, 0.51; 95% CI, .21–.82; HPV-18: 0.65; .08–1.22) than those that used the SPF10-DEIA PCR method (HPV-16: 0.74; .58–.89; HPV-18: 0.84; .57–1.11). Likewise, in analyses restricted to female subjects, the point estimates were slightly but nonsignificantly stronger in those using a 6-month persistent HPV DNA end point (HPV-16: RR, 0.59; 95% CI, .41–.77; HPV-18: 0.74; .44–1.03) than in those using a one-time detection end point (HPV-16: 0.69; .59–.78; HPV-18: 0.86; .65–1.07). Finally, results were also nonsignificantly stronger in studies restricted to female subjects that reported adjusted analyses (HPV-16: RR, 0.48; 95% CI, .29–.66; HPV-18: 0.60; .23–.97) than in those that only reported unadjusted analyses (HPV-16: 0.73; .56–.91; HPV-18: 0.77; .36–1.18).

Figure 5.

Forest plot of the relative risks for human papillomavirus (HPV) type 16 or18 infection comparing HPV-16/18–seropositive with HPV-16/18–seronegative individuals, restricted to studies using assays that measure only neutralizing antibodies. Abbreviations: CI, confidence interval; ES, effect size.

DISCUSSION

This systematic review and meta-analysis representing >24 000 individuals finds that HPV antibodies acquired through natural infection provide protection against subsequent type-specific genital HPV infection. Given that the observed natural immunity is modest and limited to female subjects, coupled with the fact that many HPV-infected individuals do not seroconvert [7, 8, 35], natural immunity is probably inferior to protection obtained from HPV vaccination for formerly infected individuals [36, 37].

There are several reasons why natural HPV immunity may not be as protective as prophylactic HPV vaccination, which has demonstrated high efficacy in formerly HPV-infected individuals [36, 37]. First, the antibody titers/levels in natural immunity are considerably lower than those observed in vaccination [38]. Indeed, a recent analysis of the Costa Rica Vaccine Trial suggested that even women receiving only 1 dose of the HPV-16/18 vaccine had >4 times higher antibody titers than naturally infected women [38]. Although the minimum antibody titer necessary for protection is currently unclear, 2 recent natural immunity studies suggested that protection may be limited to individuals with relatively higher naturally acquired antibody titers [6, 11].

Second, it is possible that misclassification in some of the assays affects the results of these studies. In particular, the VLP-ELISA may be prone to misclassification, given that it measures both neutralizing and nonneutralizing antibodies at low levels and has the potential for nonspecific cross-reactivity with other agents. This potential misclassification would attenuate the natural immunity association toward the null. Indeed, in subanalyses restricted to studies using serologic assays that measure only neutralizing antibodies, we observe nonsignificantly stronger protection. However, these serologic assays measuring only neutralizing antibodies are less sensitive measures for cumulative HPV infection [9, 24, 27].

Third, it is possible that the HPV infections observed in these studies could be a mix of newly acquired and reactivated latent infection [39]. Although current PCR-based assay methods cannot distinguish between a potential new acquisition versus a reactivation, a recent study did suggest natural HPV-16 immunity among those with new sex partners (RR, 0.45; 95% CI, .23–.87), but not among those without a new sex partner during follow-up (1.08; .65–1.79), supporting the possible idea that natural immunity may protect against new acquisition but not reactivation [14]. Protection against new HPV infections is thought to be largely antibody mediated, whereas control of existing infections is probably more cell mediated. Although the strength of cell-mediated and humoral immunity against HPV may be correlated, it is possible that antibodies measured in these studies mainly show protection against HPV reacquisition but not reactivation. However, given that the seropositive individuals included in these studies may have already “cleared” an HPV infection before the start of follow-up, their cell-mediated immune responses may also differ from that of seronegative individuals. Therefore, it is conceivable that the protection observed via natural antibodies in serum may only be a surrogate for other local mechanisms that need further elucidation.

This systematic review and meta-analysis suggests that HPV natural immunity may be restricted to female subjects. Men are known to have a considerably lower HPV seroprevalence than women [40, 41], and a recent study suggested that HPV-seropositive women may have higher antibody levels than HPV-seropositive men [26]. This may be because HPV infections in women have more access to the mucosal immune system than HPV infections on the keratinized surface of the male genitals or because HPV detection in men may be more likely to represent deposition instead of infection. However, only 3 relatively modest-sized studies in male subjects measured HPV-16 antibody levels (in assays that measured both neutralizing and nonneutralizing antibodies), and they were performed separately from the studies in female subjects. Further examination of natural immunity in men is necessary, as well as examination of whether naturally acquired antibodies may protect against extracervical HPV infection, such as oral and anal HPV infection [20, 26], which are known to cause an increasing number of oropharyngeal and anal cancers in higher-income countries [42, 43].

Most of the studies in female subjects that suggested natural immunity were limited to one-time detection of cervical HPV infection, and they followed up the women for ≤4 years. In addition, none of the studies measured the date of the initial seroconversion, so it is unclear how long antibodies were present before each study's initiation. Therefore, the duration of natural immunity is still unclear, as well as whether this protection can also be extended against cervical precancer. This meta-analysis suggested nonsignificantly stronger natural immunity when restricted to studies examining HPV persistence as the outcome, and findings of one of the larger studies suggested significant natural immunity against atypical cells of undetermined significance related to HPV-16 and HPV-18 and nonsignificant protection against cervical intraepithelial neoplasia 2+ related to HPV-16 [6]. However, the number of precancers in this study was small, and further examination of the risk of precancer among “HPV-reinfected” individuals is necessary.

Understanding the risk of HPV reinfection and subsequent disease is important, given that genital HPV infection is commonly acquired soon after sexual debut [44] and the infection is cleared or controlled in most individuals [2, 3]. Therefore, the number of formerly infected individuals (that may have natural immunity) probably represents a significant proportion of the sexual active population. Indeed, a prevaccine analysis from the US National Health and Nutrition Examination Survey suggested that 20% of women and 9% of men aged ≥20 years were HPV-16 seropositive by VLP-ELISA [41], and subsequent publications suggested that only 5% of women [45] and 6% of men [46] had active HPV-16 infection.

Given that this analysis suggested only a modest effect of natural immunity restricted to formerly infected women, it suggests that HPV natural immunity may not have a strong impact on the declining cervical HPV prevalence by age in certain countries [10] or in the reduced cost-effectiveness of HPV vaccination of older individuals [47]. However, the age of participants varied considerably across the different studies included in this meta-analysis, and we were unable to adequately evaluate potential effect modification of natural immunity by age, given the lack of individual person-level data. So, although the lower cervical HPV prevalence at older ages may be more due to a reduced number of new sexual partners with age, we cannot exclude a significant impact of natural immunity in certain age ranges [48]. Further study is needed to confirm whether the reduction in sexual partners at older ages has a larger impact on the cost-effectiveness of HPV vaccination in older individuals than the observed natural immunity.

Like other systematic reviews and meta-analysis, this study is subject to limitations. Our analysis was unable to examine potential sources of heterogeneity outside of sex, such as age, tobacco use, and HIV status, which could affect the results. We also used studies that had different statistical methods to calculate their effect estimates, including many studies that did not attempt to adjust for any potential confounders. However, results seemed fairly similar when comparing studies that presenting adjusted estimates with those that used unadjusted estimates, although unadjusted estimates may have modestly attenuated to the null. This analysis was also limited in examining natural immunity against types other than HPV-16 and HPV-18. Although our natural immunity estimates seemed similar by type, there were few existing studies examining non–HPV-16/18 types. Our analyses were restricted to genotype-specific immunity, because further research is necessary to determine whether there may be any natural immunity against phylogenetically related HPV types (ie, “cross-protection”) like what is observed with HPV vaccination [49].

This is the first systematic review and meta-analysis to examine natural HPV immunity in both female and male subjects. The modest natural immunity observed in this study suggests that many previously exposed unvaccinated individuals are still at risk for future HPV infection and that natural immunity is probably inferior to HPV vaccination in protecting against reinfection.

Supplementary Data

Supplementary materials are available at http://jid.oxfordjournals.org. Consisting of data provided by the author to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the author, so questions or comments should be addressed to the author.