Abstract
Background
People living with HIV (PLWH) are more likely to smoke and harbor oral human papillomavirus (HPV) infections, putting them at higher risk for head and neck cancer. We investigated effects of HIV and smoking on oral HPV risk.
Methods
Consecutive PLWH (n = 169) and at-risk HIV-negative individuals (n = 126) were recruited from 2 US health centers. Smoking history was collected using questionnaires. Participants provided oral rinse samples for HPV genotyping. We used multivariable logistic regression models with interaction terms for HIV to test for smoking effect on oral HPV.
Results
PLWH were more likely to harbor oral HPV than HIV-negative individuals, including α (39% vs 28%), β (73% vs 63%), and γ-types (33% vs 20%). HIV infection positively modified the association between smoking and high-risk oral HPV: odds ratios for smoking 3.46 (95% confidence interval [CI], 1.01–11.94) and 1.59 (95% CI, .32–8.73) among PLWH and HIV-negative individuals, respectively, and relative excess risk due to interaction (RERI) 3.34 (95% CI, −1.51 to 8.18). RERI for HPV 16 was 1.79 (95% CI, −2.57 to 6.16) and 2.78 for β1-HPV (95% CI, −.08 to 5.65).
Conclusion
Results show tobacco smoking as a risk factor for oral HPV among PLWH.
Keywords: HIV, tobacco, HPV, smoking
Our finding that HIV infection modifies the relationship between tobacco smoking, oral HPV infection, and oral cancers is important, as this may have public health implications if high-risk PLWH can be targeted for cancer screening and smoking cessation interventions.
Human papillomavirus (HPV) infection is an established risk factor for a distinct subset of head and neck cancers (HNCs), accounting for an increasing share of incident HNCs in North America, including particular cancers originating in the oropharyngeal palatine and lingual tonsils [1, 2]. Three species groups of HPV are well known to infect humans: alpha (α), beta (β), and gamma (γ), with the α-HPV types being most extensively studied due to their established associations with cervical and anogenital diseases [3]. Although the α-HPV types (overwhelmingly HPV 16) are responsible for over 90% of HPV-positive HNCs, recent evidence suggests an etiologic role for types that were until recently not known to infect the oral cavity [4], including β (HPV 5 and 8) and γ (γ-11 and γ-12) types [5, 6].
People living with human immunodeficiency virus (PLWH) have a higher prevalence of all types of oral HPV [4, 7], and the molecular and mechanistic interactions that underpin this relationship have been previously described [8]. As with most cancers, PLWH are also at higher risk of HPV-related HNCs [9]. Tobacco smoking, a strong and independent risk factor for HNCs prevalent among PLWH [10], is also an important risk factor for oral HPV infection [11–13]. However, this association has only been demonstrated for the detection of α-HPV types (especially high-risk types) known to infect the mucosal epithelia of the anogenital tract. Furthermore, whereas differences in prevalence and risk of oral HPV in PLWH compared to HIV-negative populations have been demonstrated, the interactive effects of HIV and smoking have not been fully explored [7].
Many studies have suggested that smoking has adverse effects on immunity, both systemic and mucosal [14, 15]. We hypothesized that the synergistic effects of HIV infection and smoking combined will translate into higher risks of oral HPV infection, and that the risk of all types of oral HPV will be greater among PLWH compared to HIV-negative individuals.
MATERIALS AND METHODS
Study Population
The Multicenter Oropharyngeal Squamous Atypical Lesion (MuCOSAL) study was a prospective study, conducted between 2004 and 2013, of HIV-positive and HIV-negative patients presenting with and without oral lesions to the infectious disease, dental, oral medicine, and otolaryngology outpatient clinics at the Montefiore Medical Center (Bronx, NY) and Rutgers School of Dental Medicine (Newark, NJ). Where possible, HIV-negative participants were frequency matched to HIV-positive patients on age (±10 years) and sex. Exclusion criteria were minors younger than 21 years and cancer treatment in the preceding year. Participants were followed-up for approximately 1 year, with 3 visits at baseline, 6 months, and 12 months. The study protocol was approved by the institutional review boards at the participating institutions and all participants provided written informed consent prior to participating.
Data Collection
The main components of the data collection process for this study have been described previously [16]. Briefly, participants underwent a structured interview administered by trained research personnel. We collected information on sociodemographic characteristics (such as age, sex, race/ethnicity, and occupation); lifestyle habits (eg, cigarette smoking and alcohol consumption); sexual practices and history (eg, lifetime number of sexual partners, practice of oral sex); and history of various sexual-transmitted diseases.
In addition, each participant underwent a comprehensive clinical oral examination. Patients were classified based on presentation of the most severe oral lesion, if detected, including benign/papilloma (eg, oral warts, frictional keratosis) and potentially malignant lesions (eg, lichen planus, leukoplakia, and/or erythroplasia) [17].
Participants provided a 30-second gargle and rinse sample at each visit using Scope mouthwash for HPV genotyping. The samples were kept on ice until stored at −20°C within 6 hours of collection and thereafter transferred to the laboratory for DNA isolation. In the current analyses, baseline questionnaire, clinical oral examination, and HPV genotyping data were used to test for the effect of smoking on HPV detection by HIV serostatus.
Smoking History and Covariates
Detailed smoking history was collected by asking participants to divide their lifetime consumption into varying periods of duration and intensity. For each period, information on age at start, age at cessation, and the number of cigarettes smoked per day in that period was collected. From this, smoking duration was calculated and pack-years computed as the number of cigarettes smoked daily divided by 20 and multiplied by the duration. Ever-smokers were defined as those who smoked tobacco for at least 1 year at any point in their lives.
Age, sex, race/ethnicity, lifetime number of sexual partners, and oral sexual history are factors that have been shown to be independently associated with oral HPV detection [2, 11, 16, 18]. Also, HPV is associated with the presence of both benign and potentially malignant lesions [19]. Therefore, we considered these variables as possible confounders in our study. Also, we considered HIV infection a direct effect modifier of the smoking-HPV relationship. We represented these factors and their relationships with the exposure (smoking) and outcome (HPV detection) in our proposed causal directed acyclic graph for the overall smoking-HPV relationship (Supplementary Figure 1), while estimating the direct effect modification by HIV infection using the approach proposed by Vanderweele and Robbins [20].
HPV DNA Testing
HPV DNA detection and typing was performed using a next-generation sequencing (NGS) assay developed in our research group [21]. DNAs extracted from the oral rinse sample and brush biopsy specimens were tested for the presence of β- and γ-HPV types using 2 different polymerase chain reaction (PCR) assays, termed FAP and E1. These assays are similar to the SPF assay, which was designed to amplify α-HPVs, as described in Fonseca et al [21]. These assays operate similarly to the 16S microbiome assays that amplify a specific fragment using bar-coded primers. For β- and γ-HPV detection, a unique 12 bp Golay DNA barcode was assigned to each subject and introduced into the PCR amplicons. Barcoded PCR products from multiple subjects were pooled at approximately equal molar DNA concentrations and sequenced on an Illumina platform. The millions of reads produced by NGS were demultiplexed with individual samples being reconstructed by sorting barcodes into unique files. The reads were then matched with known or novel HPV sequences using BLAST, and a specialized HPV database developed by our laboratory. Positive and negative controls were included with each PCR, as well as samples previously identified to have a representation of different β- and γ-HPV types. Identification of individual HPV types was based on parameters including total read number, percent of total HPV reads, and distribution of reads for each HPV type.
For the study outcomes, we classified HPV phylogenetically into α-, β-, and γ-HPV genera and the relevant species corresponding to these genera: α-3, α-9; β-1, β-2, β-3; γ-7, γ-8, γ-9 [3]. Furthermore, we grouped together α-HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59 as high-risk α-types following the International Agency for Research on Cancer classification [6], and we singled out α-HPV 16, due to its predominant role in HNCs, as well as the recently implicated β-1 species types, HPV 5 and HPV 8 [5, 6].
Statistical Analyses
Unconditional logistic regression modelling was used to estimate the odds ratio (OR) and confidence interval (CI) for the smoking-HPV relationship while adjusting for previously specified confounders. Smoking was modelled following the recommendation by Leffondré et al [22]. First, we included a centered continuous smoking pack-years variable by deducting the mean pack-years from each ever-smoker while keeping the value for never-smokers as zero. Then we added an indicator variable for never and ever smoker. Modelling smoking this way accounted for the quantitative (duration) and qualitative (intensity) differences between a never and ever smoker, while allowing for the interpretation of the coefficient of the indicator variable as the effect of average smoking pack-years among ever-smokers on oral HPV detection compared to never-smokers. This transformation and modelling strategy provides a superior model fit, aids interpretation, and does not alter the regression coefficients [22, 23].
We assessed for effect modification based on an epidemiological definition: that the effect of 1 variable (smoking, the exposure) on another (oral HPV, the outcome) varies across the strata of a third (HIV, the effect modifier) [20], by including an interaction term for HIV in the logistic model, and reporting this following the 4 steps recommended by Knol and Vanderweele [24]. This included estimating ORs for each stratum of HIV and smoking and their joint effects, alongside the effect of smoking on HPV within the strata of HIV and the relative excess risk due to interaction (RERI).
RERI represents the difference between the joint OR and the individual contributions by smoking and HIV, calculated as: OR11 − OR01 − OR10 + 1; where a RERI of 0 indicates perfect additivity (ie, no effect modification), and a value of greater or less than 0 indicates positive or negative additive effect modification, respectively [25]. In addition, we estimated the attributable proportion (AP) of oral HPV risk associated with both smoking and HIV, which was calculated as: RERI ÷ OR11 [25]. For the few outcomes where either of the exposures was protective (ie, either OR01 or OR10 was < 1), the exposure was recoded as a risk factor in order to estimate the correct RERI and AP [26]. Confidence intervals and P values were computed from the standard error estimates using the delta method [27].
As sensitivity analyses, we modelled smoking simply as never, past, and current, and retested the effect modification by HIV for high-risk for α-HPV, HPV 16, and β-1. All analyses were conducted in R Studio (V 1.2.1335) using the R statistical programming language [28].
RESULTS
A total of 325 participants were recruited into the study. HPV genotyping results were missing for 30 subjects, leaving 295 adults as the final sample size in this analysis, comprising 126 HIV-negative and 169 HIV-positive individuals. The mean age of the study participants was 52.1 years (±11.8 SD), with approximately half being male (51.2%) and half identifying as African-American (49.1%). Table 1 further describes the characteristics of the study participants stratified by HIV serostatus. HIV-positive participants were younger on average, had a higher number of lifetime sexual partners, smoked more cigarettes, and drank more alcohol across their lifetimes, but were less likely to have oral lesions. Among HIV-negative individuals, prevalence of any α-HPV or β-1 HPV was associated with the presence of benign/papilloma or precursor/dysplastic lesions, respectively (Supplementary Table 2); there were no significant differences in oral HPV prevalence by oral lesion status among PLWH (Supplementary Table 3).
Table 1.
Selected Characteristics of Study Participants by HIV Status
| Characteristic | HIV-Negative n = 126 | HIV-Positive n = 169 | P Value |
|---|---|---|---|
| Age, mean (SD) | 52.25 (14.10) | 49.83 (8.76) | .071 |
| Sex | |||
| Female | 65 (51.6) | 79 (46.7) | .481 |
| Male | 61 (48.4) | 90 (53.3) | |
| Race | |||
| Non-Hispanic white | 24 (19.0) | 15 (8.9) | .071 |
| Hispanic | 39 (31.0) | 60 (35.5) | |
| African-American | 57 (45.2) | 88 (52.1) | |
| Other | 6 (4.8) | 6 (3.6) | |
| Ever-smoker | |||
| Never | 44 (34.9) | 40 (23.7) | .047 |
| Ever | 82 (65.1) | 129 (76.3) | |
| Smoking pack-years among ever-smokers, mean (SD) | 20.50 (22.84) | 20.17 (20.47) | .914 |
| Smoking history | |||
| Never | 44 (34.9) | 40 (23.7) | .089 |
| Current | 50 (39.7) | 84 (49.7) | |
| Past | 32 (25.4) | 45 (26.6) | |
| Lifetime ethanol use in liters, median (IQR) | 17 (0.0–158.5) | 30 (0.0–336.0) | .196 |
| Lifetime ethanol use categories | |||
| Up to 10 L | 61 (48.4) | 69 (40.8) | .422 |
| >10–124 L | 28 (22.2) | 39 (23.1) | |
| >124–369 L | 15 (11.9) | 19 (11.2) | |
| >369 L | 22 (17.5) | 42 (24.9) | |
| Type of oral lesion | |||
| None | 79 (62.7) | 132 (78.1) | .011 |
| Benign/papilloma | 19 (15.1) | 18 (10.7) | |
| Precursor/dysplastic | 28 (22.2) | 19 (11.2) | |
| Lifetime number of sexual partners | |||
| ≤5 | 61 (48.4) | 50 (29.6) | <.001 |
| 6–20 | 45 (35.7) | 54 (32.0) | |
| ≥ 21 | 20 (15.9) | 61 (36.1) | |
| NA | 0 (0.0) | 4 (2.4) | |
| Sexual preferences | |||
| WSM | 61 (48.4) | 67 (39.6) | NA |
| WSMW | 3 (2.4) | 11 (6.5) | |
| MSM | 0 (0.0) | 22 (13.0) | |
| MSWM | 57 (45.2) | 64 (37.9) | |
| NA | 5 (4.0) | 5 (3.0) | |
| Oral sex history | |||
| No | 36 (28.6) | 27 (16.0) | .024 |
| Yes | 90 (71.4) | 141 (83.4) | |
| NA | 0 (0.0) | 1 (0.6) | |
| CD4 count | |||
| 500 + T cells/mm3 | … | 50 (29.96) | … |
| >200 to < 500 T cells/mm3 | … | 54 (32.0) | |
| ≤200 T cells/mm3 | … | 17 (10.1) | |
| NA | 48 (28.4) | ||
| HIV viral load | … | ||
| <4000 copies/mL | … | 130 (76.9) | |
| 4000+ copies/mL | … | 13 (7.7) | |
| NA | … | 26 (15.4) |
Data are mean (%) except where indicated.
Abbreviations: HIV, human immunodeficiency virus; IQR, interquartile range; MSM, men who have sex with men only; MSWM, men who have sex with women only or with women and men; NA, not available; WSM, women who have sex with men only; WSWM, women who have sex with women only or with women and men.
Figure 1 shows the distribution of oral HPV types and genera detected by HIV status. Consistently, the prevalence of HPV was higher among HIV-positive participants for all genera and types, except α-14 types. In addition, no α-11, γ-3, or high-risk α-HPV 31 or 45 types were detected in oral rinse samples of HIV-negative participants.
Figure 1.
Unadjusted prevalence of oral HPV (genus/species/types) by HIV status.
Among HIV-positive participants, an ever-smoker with 15.4 pack-years (reflecting an average number of pack-years in our study population) had 3.46 times higher odds of presenting with a high-risk α-HPV type (95% CI, 1.01–11.94) compared to a never-smoker HIV-positive participant after adjusting for confounders. Increased associations were also observed for HPV 16 alone (Table 2) and β-1 HPV species (Table 3).
Table 2.
Modification of the Effect of Smoking on Oral α-HPV (Genus/Species/Types) by HIV
| HPV | Never Smoker | Ever Smoker | Effect of Smoking on Oral HPV Within Strata of HIV |
|---|---|---|---|
| Any α-HPV | |||
| HIV negative, OR | 1 | 1.67 (.68–4.32) | 1.67 (.68–4.32) |
| HIV positive, OR | 1.32 (.47–3.76) | 3.05 (1.29–7.73) | 2.32 (.96–5.56) |
| RERI | … | 1.06 (−.81 to 2.95) P = .13 | … |
| AP | … | 0.35 (−.21 to .91) P = .11 | … |
| α-3 | |||
| HIV negative, OR | 1 | 1.49 (.42–6.23) | 1.49 (.42–6.23) |
| HIV positive, OR | 1.26 (.27–5.93) | 1.61 (.48–6.58) | 1.28 (.36–4.59) |
| RERI | … | −.14 (−2.50 to 2.22) P = .54 | … |
| AP | … | −.09 (−1.53 to 1.36) P = .45 | … |
| α-9 | |||
| HIV negative, OR | 1 | 0.96 (.24–4.21) | 0.96 (.24–4.21) |
| HIV positive, OR | 2.03 (.51–8.88) | 3.32 (1.01–13.54) | 1.64 (.53–5.03) |
| RERI | … | −1.40 (−5.71 to 2.91) P = .73a | … |
| AP | … | −.66 (−2.58 to 1.26) P = .24a | … |
| α-high-risk HPVb | |||
| HIV negative, OR | 1 | 1.56 (.39–7.86) | 1.56 (.39–7.86) |
| HIV positive, OR | 1.59 (.32–8.73) | 5.48 (1.58–26.23) | 3.46 (1.01–11.94) |
| RERI | … | 3.34 (−1.51 to 8.18) P = .08 | … |
| AP | … | 0.61 (.12–1.10) P = .007 | … |
| α-HPV 16 | |||
| HIV negative, OR | 1 | 1.60 (.30–12.24) | 1.60 (.30–12.24) |
| HIV positive, OR | 1.75 (.26–14.41) | 4.14 (.89–30.93) | 2.37 (.53–10.54) |
| RERI | … | 1.79 (−2.57 to 6.16) P = .20 | … |
| AP | … | 0.43 (−.40 to 1.26) P = .15 | … |
ORs adjusted for age, sex, race, smoking pack-years, oral sex history, and lifetime number of sexual partners. Values in parentheses are 95% confidence intervals.
Abbreviations: AP, attributable proportion; HIV, human immunodeficiency virus; HPV, human papillomavirus; OR, odds ratio; RERI, relative excess risk due to interaction.
aAt least 1 exposure was preventive and so it was recoded as a risk factor to correctly estimate RERI and AP.
bHPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59.
Table 3.
Modification of the Effect of Smoking on Oral β-HPV (Genus/Species/Types) by HIV
| HPV | Never Smoker | Ever Smoker | Effect of Smoking on Oral HPV Within Strata of HIV |
|---|---|---|---|
| Any β-HPV | |||
| HIV negative, OR | 1 | 0.87 (.37–2.02) | 0.87 (.37–2.02) |
| HIV positive, OR | 1.00 (.39–2.55) | 2.50 (1.07–5.83) | 2.49 (1.04–5.98) |
| RERI | … | −1.85 (−4.64 to .93) P = .90a | … |
| AP | … | −1.62 (−3.92 to .68) P = .08a | … |
| β-1 | |||
| HIV negative, OR | 1 | 1.12 (.49–2.60) | 1.12 (.49–2.60) |
| HIV positive, OR | 1.70 (.67–4.39) | 4.60 (2.03–10.86) | 2.70 (1.16–6.28) |
| RERI | … | 2.78 (−.08 to 5.65) P = .02 | … |
| AP | … | 0.60 (.24–.96) P = .0005 | … |
| β-2 | |||
| HIV negative, OR | 1 | 0.94 (.42–2.15) | 0.94 (.42–2.15) |
| HIV positive, OR | 0.80 (.31–2.02) | 1.27 (.58–2.87) | 1.60 (.69–3.69) |
| RERI | … | −0.67 (−2.31 to .97) P = .78a | … |
| AP | … | −0.57 (−1.87 to .73) P = .19a | … |
| β-3 | |||
| HIV negative, OR | 1 | 0.31 (.09–1.08) | 0.31 (.09–1.08) |
| HIV positive, OR | 0.54 (.13–2.05) | 0.42 (.13–1.39) | 0.78 (.21–2.91) |
| RERI | … | −1.83 (−5.91 to 2.25) P = .81a | … |
| AP | … | −1.05 (−3.34 to 1.23) P = .18a | … |
| HPV 5 | |||
| HIV negative, OR | 1 | 0.75 (.26–2.32) | 0.75 (.26–2.32) |
| HIV positive, OR | 1.97 (.65–6.30) | 1.05 (.39–3.09) | 0.53 (.20–1.39) |
| RERI | … | 0.89 (−1.04 to 2.83) P = .18a | … |
| AP | … | 0.34 (−.35 to 1.03) P = .16 | … |
| HPV 8 | |||
| HIV negative, OR | 1 | 1.14 (.26–6.03) | 1.14 (.26–6.03) |
| HIV positive, OR | 1.78 (.36–9.95) | 2.93 (.79–14.79) | 1.64 (.46–5.92) |
| RERI | … | 1.00 (−1.93 to 3.94) P = .25 | … |
| AP | … | 0.34 (−.59 to 1.28) P = .23 | … |
ORs adjusted for age, sex, race, smoking pack-years, oral sex history and lifetime number of sexual partners. Values in parentheses are 95% confidence intervals.
Abbreviations: AP, attributable proportion; HIV, human immunodeficiency virus; HPV, human papillomavirus; OR, odds ratio; RERI, relative excess risk due to interaction.
aAt least 1 exposure was preventive and so it was recoded as a risk factor to correctly estimate RERI and AP.
For α-genus HPV, the joint effect estimates (ie, ever smoker + HIV positive, OR11) for all the outcomes were consistently larger than the individual effects (ie, ever smoker alone, OR01; and HIV positive alone, OR10) (Table 2). On the additive scale, HIV positively modified the effect of smoking on the detection of oral α-HPV types, including HPV 16, with the strongest measures of effect modification observed for high-risk α-types (RERI = 3.34 [95% CI, 1.51–8.18] and AP = 0.61 [95% CI, .12–1.10]) (Table 2). The results indicated a negative effect modification for the detection of α-3 and α-9 species types (except HPV 16).
HIV also showed superadditive (positive) effect modification on the associations between smoking and oral β-HPV types HPV 8 and 5, and β-1 species types, but negative effect modification for β-2 and β-3 types (Table 3), as well as with γ-8 and γ-10 species types (Supplementary Table 1). Overall, trend estimates indicated stronger smoking-HPV associations among PLWH compared to HIV-negative individuals (right-most columns of Table 2, Table 3, and Table 4), consistent with our hypothesis of effect modification by HIV.
Table 4.
Sensitivity Analyses for the Modification of the Effect of Smoking on Oral HPV (α-High Risk, HPV 16, and β-1 Species) by HIV
| HPV | Smoking | Past Smoker on Oral HPV Within Strata of HIV | Current Smoker on Oral HPV Within Strata of HIV | ||
|---|---|---|---|---|---|
| Never | Past | Current | |||
| α-High-risk HPVa | |||||
| HIV negative, OR | 1.0 | 1.52 (.24–9.69) | 1.53 (.32–8.53) | 1.52 (.24–9.69) | 1.53 (.32–8.53) |
| HIV positive, OR | 1.58 (.32–8.72) | 4.33 (1.05–22.84) | 6.24 (1.72–30.76) | 3.33 (.85–13.10) | 3.55 (.98–12.80) |
| RERI | … | 2.23 (−2.53 to 6.99) P = .17 | 4.12 (−2.03 to 10.28) P = .09 | … | … |
| AP | … | 0.51 (−.25 to 1.28) P = .09 | 0.66 (.19–1.13) P = .002 | … | … |
| α-HPV 16 | |||||
| HIV negative, OR | 1.0 | 1.45 (.15–14.19) | 1.69 (.24–14.70) | 1.45 (.15–14.19) | 1.69 (.24–14.70) |
| HIV positive, OR | 1.76 (.26–14.53) | 3.29 (.55–27.48) | 4.85 (.96–38.10) | 2.12 (.40–11.30) | 2.54 (.53–12.14) |
| RERI | … | 1.08 (−3.73 to 5.90) P = .33 | 2.41 (−3.26 to 8.07) P = .20 | … | … |
| AP | … | 0.33 (−.94 to 1.60) P = .31 | 0.50 (−.30 to 1.29) P = .11 | … | … |
| β-1b | |||||
| HIV negative, OR | 1.0 | 1.21 (.43–3.47) | 1.06 (.42–2.69) | 1.21 (.43–3.47) | 1.06 (.42–2.69) |
| HIV positive, OR | 1.70 (.67–4.38) | 3.85 (1.47–10.55) | 5.03 (2.11–12.53) | 2.29 (.86–6.15) | 2.93 (1.20–7.14) |
| RERI | … | 1.95 (−1.44 to 5.34) P = .13 | 3.27 (−.24 to 6.78) P = .03 | … | … |
| AP | … | 0.51 (−.11 to 1.12) P = .05 | 0.65 (.30–1.00) P = .0001 | … | … |
ORs adjusted for age, sex, race, oral sex history, and lifetime number of sexual partners. Values in parentheses are 95% confidence intervals.
Abbreviations: AP, attributable proportion; HIV, human immunodeficiency virus; HPV, human papillomavirus; OR, odds ratio; RERI, relative excess risk due to interaction.
aHPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59.
bIncludes HPV 5 and 8.
In the sensitivity analyses, where smoking history was modelled simply as never, past, and current (Table 4), the joint effect estimates for current smoker PLWH were higher than for HIV-positive past smokers, for high-risk α-HPV types (including HPV 16) and β-1 species types (Table 4). These dose-response associations were also reflected by the measures of additive effect modification (ie, RERI and AP for each outcome).
DISCUSSION
In this study, we examined how the association between smoking and oral HPV is modified by HIV. We show positive effect modification by HIV on the smoking for high-risk α-HPV types, including HPV 16 alone, and for β-1 species types, with a high attributable proportion for HPV detection in the oral cavity associated with exposure to both smoking and HIV. In addition, we found a higher (unadjusted) prevalence for almost all α-HPV types in PLWH compared to HIV-negative individuals, which is consistent with the literature [7, 16, 29, 30]. In addition, we found higher prevalence of β- and γ-HPV types among PLWH, which was novel. We also report a higher prevalence of oral lesions among the HIV-negative individuals, despite their lower oral HPV prevalence. This was likely due to recruiting HIV-negative individuals from dental and oral medicine clinics at the early stage of the study, which may have oversampled patients with oral lesions; importantly, however, all our reported effect estimates were adjusted for each individual’s lesion status (none, benign/papilloma, or precursor/dysplastic).
The immunosuppressive effects of cigarette smoke have been well described [31, 32], including adverse effects on both systemic and mucosal immunity [14, 15]. The combination of tobacco-induced and HIV-induced immunosuppression may explain the positive RERI and high AP for oral HPV detection observed in our study. This suggests that PLWH who smoke may be at higher risk of HPV-associated HNCs [25].
Deployment of oral HPV testing/screening for HNC prevention will be challenging, due to the low prevalence of high-risk oral HPV in the general population [33]. Consequently, for oral HPV screening to be viable and useful, there needs to be identification of high-risk subgroups [33, 34], and there is a growing public health interest in identifying these populations [35, 36]. Our study findings suggest that PLWH who smoke may be such a candidate group.
To the best of our knowledge, our study is the first to show effect modification by HIV of the smoking-oral HPV relationship. A previous study by Beachler et al [30] found no relationship between smoking intensity and duration, and detection of α-type oral HPV in both HIV-positive and negative individuals. This contrasts with our finding of an association between smoking and α-high-risk oral HPV among PLWH but not in HIV-negative individuals, and the observed risk disparity with β-1 oral HPV. The differing results could be due to how the 2 studies modelled smoking history and/or differences in HPV DNA testing methods; in their study, Beachler et al [30] tested for α-types by PGMY09/11 PCR, whereas we employed NGS. However, their smoking modelling strategy was not described.
There are a few limitations to consider in the interpretation of our results. First, because our analyses were based on the baseline data (ie, were cross-sectional), the temporal relationship between smoking, HPV, and HIV infection is unclear. This notwithstanding, our conclusions are plausible, given that the effects of smoking on oral HPV has been shown prospectively and with increasing smoking pack-years [12], and smoking has been shown to be associated with a reduced clearance rate of oral HPV [7]. Second, the NGS method we used to test for HPV is a highly sensitive assay with elevated risk for false positives; this, combined with recall error inherent in measuring lifetime cigarette smoking and other self-reported covariates, increases the potential for misclassification bias in our results.
In conclusion, while previous studies have shown an association between smoking and detection of α-HPV in the oral cavity, this study was the first to show associations for non-α-HPV types, and to provide evidence of effect modification by HIV on the relationship between smoking and oral HPV, with a significant attributable proportion of high-risk HPV associated with combined exposure to smoking and HIV. Although we have previously reported a role in HNC risk for detection of several β- and γ-HPV types in the oral cavity [5], further evidence is needed from studies of HIV-infected populations, who are at higher risk of HNC [9]. Also, importantly, we showed that PLWH, with an average of 15.4 smoking pack-years, are at higher risk of presenting with oral HPV compared to HIV-positive never-smokers. If our findings are confirmed, this may have public health implications if high-risk PLWH can be targeted for preferential HNC screening and smoking cessation interventions.
Supplementary Data
Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Notes
Financial support. This work was supported by the National Institute of Dental and Craniofacial Research (grant numbers DE021671 to N. F. .S. and DE026177 to N. F. .S. and R. D. B.); National Cancer Institute (grant numbers CA013330 to the Einstein Cancer Research Center and CA016056 to the Roswell Park Comprehensive Cancer Center); B. Y. A. was funded by a McGill University International Graduate Mobility Award and a Réseau de Recherche en Santé Buccodentaire et Osseuse fellowship.
Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
References
- 1. Laprise C, Madathil SA, Schlecht NF, et al. Human papillomavirus genotypes and risk of head and neck cancers: results from the HeNCe Life case-control study. Oral Oncol 2017; 69:56–61. [DOI] [PubMed] [Google Scholar]
- 2. Chaturvedi AK, Engels EA, Pfeiffer RM, et al. Human papillomavirus and rising oropharyngeal cancer incidence in the United States. J Clin Oncol 2011; 29:4294–301. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. de Villiers EM, Fauquet C, Broker TR, Bernard HU, zur Hausen H. Classification of papillomaviruses. Virology 2004; 324:17–27. [DOI] [PubMed] [Google Scholar]
- 4. Bottalico D, Chen Z, Dunne A, et al. The oral cavity contains abundant known and novel human papillomaviruses from the Betapapillomavirus and Gammapapillomavirus genera. J Infect Dis 2011; 204:787–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Agalliu I, Gapstur S, Chen Z, et al. Associations of oral α-, β-, and γ-human papillomavirus types with risk of incident head and neck cancer. JAMA Oncol 2016; 2:599–606. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Bouvard V, Baan R, Straif K, et al. ; WHO International Agency for Research on Cancer Monograph Working Group A review of human carcinogens–Part B: biological agents. Lancet Oncol 2009; 10:321–2. [DOI] [PubMed] [Google Scholar]
- 7. Beachler DC, Sugar EA, Margolick JB, et al. Risk factors for acquisition and clearance of oral human papillomavirus infection among HIV-infected and HIV-uninfected adults. Am J Epidemiol 2015; 181:40–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Verma M, Erwin S, Abedi V, et al. Modeling the mechanisms by which HIV-associated immunosuppression influences HPV persistence at the oral mucosa. PLoS One 2017; 12:e0168133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Beachler DC, DʼSouza G. Oral human papillomavirus infection and head and neck cancers in HIV-infected individuals. Curr Opin Oncol 2013; 25:503–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Kirk GD, Merlo CA; Lung HIV Study HIV infection in the etiology of lung cancer: confounding, causality, and consequences. Proc Am Thorac Soc 2011; 8:326–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Fakhry C, Gillison ML, D’Souza G. Tobacco use and oral HPV-16 infection. JAMA 2014; 312:1465–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Kero K, Rautava J, Syrjänen K, Willberg J, Grenman S, Syrjänen S. Smoking increases oral HPV persistence among men: 7-year follow-up study. Eur J Clin Microbiol Infect Dis 2014; 33:123–33. [DOI] [PubMed] [Google Scholar]
- 13. Gaester K, Fonseca LA, Luiz O, et al. Human papillomavirus infection in oral fluids of HIV-1-positive men: prevalence and risk factors. Sci Rep 2014; 4:6592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Sopori M. Effects of cigarette smoke on the immune system. Nat Rev Immunol 2002; 2:372–7. [DOI] [PubMed] [Google Scholar]
- 15. Calvo M, Laguno M, Martínez M, Martínez E. Effects of tobacco smoking on HIV-infected individuals. AIDS Rev 2015; 17:47–55. [PubMed] [Google Scholar]
- 16. Muller K, Kazimiroff J, Fatahzadeh M, et al. Oral human papillomavirus infection and oral lesions in HIV-positive and HIV-negative dental patients. J Infect Dis 2015; 212:760–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Van Der Waal I. Oral leukoplakia: present views on diagnosis, management, communication with patients, and research. Curr Oral Health Rep 2019; 6:9–13. [Google Scholar]
- 18. Gillison ML, Broutian T, Pickard RK, et al. Prevalence of oral HPV infection in the United States, 2009–2010. JAMA 2012; 307:693–703. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Shiels MS, Althoff KN, Pfeiffer RM, et al. ; North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD) of the International Epidemiologic Databases to Evaluate AIDS (IeDEA) HIV infection, immunosuppression, and age at diagnosis of non-AIDS-defining cancers. Clin Infect Dis 2017; 64:468–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. VanderWeele TJ, Robins JM. Four types of effect modification: a classification based on directed acyclic graphs. Epidemiology 2007; 18:561–8. [DOI] [PubMed] [Google Scholar]
- 21. Fonseca AJ, Taeko D, Chaves TA, et al. HPV infection and cervical screening in socially isolated indigenous women inhabitants of the Amazonian rainforest. PLoS One 2015; 10:e0133635. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Leffondré K, Abrahamowicz M, Siemiatycki J, Rachet B. Modeling smoking history: a comparison of different approaches. Am J Epidemiol 2002; 156:813–23. [DOI] [PubMed] [Google Scholar]
- 23.Kleinbaum DG, Kupper LL, Nizam A, Rosenberg ES. Applied regression analysis and other multivariable methods. 5th ed. Cengage Learning, MA; 2014. [Google Scholar]
- 24. Knol MJ, VanderWeele TJ. Recommendations for presenting analyses of effect modification and interaction. Int J Epidemiol 2012; 41:514–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. VanderWeele TJ, Knol MJ. A tutorial on interaction. Epidemiol Methods 2014; 3:33–72. [Google Scholar]
- 26. Knol MJ, VanderWeele TJ, Groenwold RH, Klungel OH, Rovers MM, Grobbee DE. Estimating measures of interaction on an additive scale for preventive exposures. Eur J Epidemiol 2011; 26:433–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Hosmer DW, Lemeshow S. Confidence interval estimation of interaction. Epidemiology 1992; 3:452–6. [DOI] [PubMed] [Google Scholar]
- 28. R Core Team. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing, 2019. [Google Scholar]
- 29. Mooij SH, Boot HJ, Speksnijder AG, et al. Oral human papillomavirus infection in HIV-negative and HIV-infected MSM. AIDS 2013; 27:2117–28. [DOI] [PubMed] [Google Scholar]
- 30. Beachler DC, Weber KM, Margolick JB, et al. Risk factors for oral HPV infection among a high prevalence population of HIV-positive and at-risk HIV-negative adults. Cancer Epidemiol Biomarkers Prev 2012; 21:122–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Hernandez CP, Morrow K, Velasco C, Wyczechowska DD, Naura AS, Rodriguez PC. Effects of cigarette smoke extract on primary activated T cells. Cell Immunol 2013; 282:38–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Desrichard A, Kuo F, Chowell D, et al. Tobacco smoking-associated alterations in the immune microenvironment of squamous cell carcinomas. J Natl Cancer Inst 2018; 110:1386–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. D’Souza G, McNeel TS, Fakhry C. Understanding personal risk of oropharyngeal cancer: risk-groups for oncogenic oral HPV infection and oropharyngeal cancer. Ann Oncol 2017; 28:3065–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Zhao M, Rosenbaum E, Carvalho AL, et al. Feasibility of quantitative PCR-based saliva rinse screening of HPV for head and neck cancer. Int J Cancer 2005; 117:605–10. [DOI] [PubMed] [Google Scholar]
- 35. Moyer VA; U.S. Preventive Services Task Force Screening for oral cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med 2014; 160:55–60. [DOI] [PubMed] [Google Scholar]
- 36. Syrjänen S. Human papillomaviruses in head and neck carcinomas. N Engl J Med 2007; 356:1993–5. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.