Genital Human Papillomavirus: DNA based Epidemiology

Genital Human Papillomavirus:

DNA based EpidemiologyAnil K.Chaturvedi, D.V.M., M.P.H

Human Papillomavirus (HPV)

• Papillomaviridae

• Most common viral STD

• Double stranded DNA virus ~8 Kb

• Entire DNA sequence known

HPV genome

Classification of HPV types

• Defined by <90% DNA sequence homology in L1, E6 and E7 genes

• >100 recognized types, at least 40 infect genital tract

• 90-98% homology- sub-types

• <2% heterogeneity- intratype variants

Genital HPV- Histo-pathology

•

*Tyring SK, American journal of medicine, 1997

HPV and Cervical cancer

• Second most common cancer worldwide

• HPV is a “ necessary cause”: ~ 99.7% of cervical cancer cases

• Support from several molecular and epidemiologic studies

• Protein products of E6 and E7 genes oncogenic

HPV-molecular biology

Tindle RW, Nature Reviews, Cancer, Vol2: Jan2002

HPV-molecular biology

Herald Zur Hausen, Nature Reviews, Cancer Volume 2:5; May; 2002.

HPV- Oncogenic transformation

HPV-Epidemiology

Koutsky, LA, American Journal of Medicine, May 5, Vol 102, 1997.

Crude estimates of HPV impact in women >15 years

Developed countries

Developing countries

HPV-DNA (%) 10 15

Genital warts (%)

1 1.5

In-situ cancer 550,0000 ??

Invasive cancer 150,0000 225,0000

Mean Survival (years)

10 5

Cervical cancer in US

0

5

10

15

20

25

30

35

Year

Rate

per 1

00,0

00

AllCaucasianAfrican-American

SEER data and Statistics, CDC.

Diagnosis

• Pap smears- Current recommendations (US)

• Normal on 3 consecutive annual- 3 year screening

• Abnormal-no HPV- Annual• Abnormal- evidence of HPV- 6-12

months• LSIL/HSIL- colposcopy

HPV diagnosis

Clinical diagnosis: Genital wartsEpithelial defects

See cellular changes caused by the virus: Pap smear screening

Directly detect the virus: DNA hybridization or PCR*

Detect previous infection: Detection of antibody against HPV*

* Done in the Hagensee Laboratory

Utility of HPV screening

• Primary prevention of CC

• Secondary prevention

• Component of Bethesda 2001 recommendations

• Prevalent genotypes for vaccine design strategies

Natural history of Cervical neoplasia

CIN I CIN II CIN III

CC

1%

5%12%

Rates of progression

HPV-CC: epidemiologic considerations

• HPV is a “necessary cause”, not a “sufficient cause” for CC

• Near perfect sensitivity P(T+/D+), very poor positive predictive value P(D+/T+)

• Interplay of co-factors in progression

Host genetic•P53 and HLA polymorphisms

Herald Zur Hausen, Nature Reviews, Cancer Volume 2:5; May; 2002

HIV+ vs. HIV- story

• HIV+ men and women, 4-6 times greater risk of incident, prevalent and persistent HPV infections

• Increased cytologic abnormalities and HPV associated lesions difficult to treat

Prevalence of 27 HPV genotypes in Women with

Diverse Profiles

Anil K Chaturvedi1, Jeanne Dumestre2, Ann M. Gaffga2, Kristina M. Mire,2Rebecca A.Clark2, Patricia S.Braly3, Kathleen Dunlap3,Patricia J.

Kissinger1, and Michael E. Hagensee2

Goals of study

1. Characterize prevalent HPV types in 3 risk settings-Low-risk HIV-, high-risk HIV- and HIV+ women

2. Characterize geotypes associated with cytologic abnormalities

3. Risk factor analyses

Methods

Low-risk clinicN=68

High-risk clinicN=376

HIV+N=167

N=611

Cervical swabs and Pap smears

N=363 Took screening questionnaire

36 LR (52.9%)232 HR (61.7%)95 HIV+ (56.8%)

Methods

• Inclusion/ exclusion criteria:

• >18 years

• Non-pregnant

• Non-menstruating

• Chronic hepatic/ renal conditions

• Informed consent

Methods

• HPV assessment:

DNA from cervical swabsPolymerase chain reaction using PGMy09/11 consensus primer system reverse line hybridization (Roche molecular systems, CA)

HPV genotyping

Roche molecular systems Inc., Alameda, CA.

HPV classification

• Strip detects 27 HPV types (18 high-risk, 9 low-risk types)

• Types 6, 11, 40, 42, 53, 54, 57, 66, 84 : low-eisk

• Types 16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 55, 56, 58, 59, 68, 82, 83, 73: high-risk

• Classified as Any HPV, HR, LR, and multiple (any combination)

Pap smears

• Classified – 1994 Bethesda recommendations

• Normal, ASCUS, SIL (LSIL and HSIL)

Data analysis

• Bivariate analyses- Chi-squared or Fischer’s exact

• Binary logistic regression for unadjusted and adjusted OR and 95% CI

• Multinomial logistic regression for Pap smear comparisons (Normal, ASCUS and SIL)

Analysis

• Risk factor analysis for HPV infection- Any, HR, LR and multiple (dependent variables)

• P<0.20 on bivariate and clinically relevant included in multivariate

• All hypothesis two-sided, alpha 0.05

• No corrections for multiple comparisons

Demographics of cohort

• HIV+ older than HIV- [34.51 (SD=9.08) vs. 26.72 (SD=8,93) ] p<0.05• Predominantly African American ~80%• HIV+ more likely to report history of STD

infections, multiparity, smoking (ever) and # of sex partners in last year ( All P<0.05)

• 16.8% of HIV+ immunosuppressed (CD4 counts < 200) • 54% Viral load >10,000 copies

Clinic comparisons

0

10

20

30

40

50

60

70

80

% P

osi

tive

HPV (+) High-Risk Low-Risk Multiple

LR clinic=68HR clinic=376HIV+=167

**

* *

* P for trend <0.001

Genotype prevalence-high-risk types

0

5

10

15

20

25

30

35

40

45

50

16 18 26 31 33 35 39 45 51 52 55 56 58 59 68

MM

4

MM

7

MM

9

# O

F SPEC

IMEN

S

Low-riskHigh-riskHIV+

Genotype prevalence-low-risk types

0

5

10

15

20

25

306 11 40 42 53 54 57 66

MM

8

# O

F SPEC

IMEN

S

Low-riskHigh-riskHIV+

Rank order by prevalenceRank Overall LR

clinicHigh-risk clinic

HIV+

1 16 16 16 83

2 83 66 52 53

3 52 53, 39 35 58, 54

Pap smear associations

• Any HPV, high-risk HPV, low-risk HPV and multiple HPV with ASCUS and LSIL (p<0.01)

• ASCUS- types 18, 35

• LSIL: 16, 35, 51, 52, 68

HIV+ sub-set analyses, N=167, multivariate  CD4 cell

counts (<200 vs.>200)

HIV-RNA viral loads

Any HPV 6.41(0.77,52.8) 2.57(0.86, 7.64)

High-risk HPV

6.42(1.34,30.8) 1.59(0.64, 3.92)

Low-risk HPV 2.79(0.99, 7.89) 2.27(0.97, 5.29)

Multiple HPV

5.92(1.85,18.8) 1.10(0.46, 2.60)

Cytologic abnormalitiesb

4.21(1.28,13.7) 0.93(0.34, 2.58)

Risk-factor analyses

• Multivariate models: simultaneous adjustment for age, prior number of pregnancies, history of STD infections (self-reported), # of sex partners in previous year and HIV status

• Any HPV: younger age (<25 years), and HIV+ status ( OR=6.31; 95%CI, 2.94-13.54)

• High-risk HPV: Younger age (<25) and HIV+ status (OR= 5.30, 2.44-11.51)

• Low-risk HPV: Only HIV status (OR=12.11, 4.04-36.26)

Conclusions

• Increased prevalence of novel/uncharacterized genotypes (83 and 53) in HIV+

• Pap smear associations on predicted patterns• CD4 counts edge viral loads out• No interaction between HPV and HIV- HPV

equally oncogenic in HIV+ and HIV-• Differential risk factor profiles for infection

with oncogenic and non-oncogenic types

Discussion

• Increased 83 and 53, also observed in HERS and WHIS reports

• Probable reactivation of latent infection

• 83 and 53 more susceptible to immune loss??- also found in renal transplant subjects

What puts HIV+ at greater risk?

Palefsky JM, Cancer epi Biomarkers and Prev, 1997.

Risk in HIV+

• 1.Increased HPV infections ?

• 2. Increased persistence ?

• 3. Systemic immunosuppression- tumor surveillance

• 4. Direct-HIV-HPV interactions?

• 5.Increased multiple infections?

Study limitations

• Cross-sectional study- no information on duration of HPV infections (big player!)

• HIV- subjects predominantly high-risk- selection bias- bias to null

• Genotypic associations based on small numbers

• Multiple comparisons- increased Type I error-chance associations

Limitations

• Incomplete demographic information- no differences in rates of HPV infections

• No associations in demographics- low power

Impact of Multiple HPV infections:

Compartmentalization of riskAnil K Chaturvedi1, Jeanne Dumestre2, Issac

V.Snowhite, Joeli A. Brinkman,2Rebecca A.Clark2, Patricia S.Braly3, Kathleen

Dunlap3,Patricia J. Kissinger1, and Michael E. Hagensee2

Background

• Multiple HPV infections- increased persistence

• Persistent HPV infection-necessary for maintenance of malignant phenotype

• Impact of multiple HPV infections- not well characterized

Goals

1.Characterize prevalence of multiple HPV infections in HIV+ and HIV- women

2. Does the risk of cytologic abnormalities differ by oncogenic-non-oncogenic combination categories

3. Compartmentalize impact of mutiple HPV infections in a multi-factorial scenario

Methods

• Cross-sectional study, non-probability convenience sample

1278 HIV-women

264 HIV+women

1542women

989 women

Cervical swabs

Both HPV and

Pap data available

Methods

• Exposure: HPV DNA status- polychotomous variable (no infection, single HPV type, HR-HR combinations, HR-LR combinations, mixed combinations)

• Exposure assessment- reverse line probe hybridization

Methods

• Outcome: Pap smear status

• Binary outcome: normal, abnormal (ASCUS and above)

Statistical analysis

• Bivariate- Chi-squared, Fischer’s exact tests

• Multivariate: Binary logistic regression, likelihood ratio improvement tests, goodness-of-fit tests (model diagnostics-best fit model)

• Covariate Adjusted attributable fractions- from best fit logistic models

Adjusted attributable fractions

• Unadjusted attributable fractions:

AF= Pr (D)- Pr (Disease/ not exposed)

Pr (Disease)• In a multi-factorial setting ??• Arrive at best-fir logistic regression model• Ln (P/1-P)= β0+β1x1+β2x2+β3x3…βnxn• Let y=β0+β1x1+β2x2+β3x3…βnxn

Adjusted attributable fractions

• Can derive predicted probability of outcome from logistic model

P= ey

1+ey

• Get predicted probability for various exposure-covariate patterns from same regression model

• Set reference levels and use original equation for estimates of adjusted attributable risks

Adjusted attributable fractions

• Cohort vs. cross-sectional situations- implications of exposure prevalences

• Can derive SE and CI

• Assumptions??

• Interpretation??

• Utility??

Results-Demographics

• HIV+ older (35.08 (SD=8.56) vs. 32.24 (SD=12.19) P<0.01

• Predominantly African American ~ 80%

Prevalence of HPV by HIV

0

5

10

15

20

25

Single MultLR

MultHR

Combo

HIV-, N=812HIV+, N=177

Prevalence of multiple HPV

02468

1012141618

1 2 3 4 5 6 7 8

# of HPV types

% , N=989

Cytology results

Normal Paps

N=655, n (%)

Abnormal paps

N= 334, n (%)

No HPV 526 (76.7) 160 (23.3)

Single type 83 (50.3) 82 (49.7)

2 low-risk types 4 (57.1) 3 (42.9)

2 high-risk types 21 (33.3) 42 (66.7)

Combination 21 (30.9) 47 (69.1)

P-for trend <0.001

Adjusted models

• Adjusted for age, and HIV status, compared to subjects with single HPV types-

Multiple high-risk types- (OR=2.08, 1.11-3.89) and LR-HR combinations ( 2.40, 1.28-4.52) risk of cytologic abnormalities

• Multiple infections linear predictor- adjusted for age and HIV, per unit increase in number (OR=1.85, 1.59, 2.15)

Adjusted attributable fractions• Possible models- Main exposure multiple

infections-No, single, multiple (Dummy variables)

Co-variates: HIV: yes, no&Age : <25 years and >=25 years

1. Intercept, HIV+, age <252. Intercept, single HPV (D1), HIV+, age < 253. Intercept, HIV-, Single HPV (D1), Multiple HPV

(D2) and age < 254. Intercept, D1, D2, HIV+, age <25

AAR

• 2 vs. 1: single HPV

• 4 vs. 2: multiple

• 4 vs. 3: HIV status

AARPercet AAR

51.89

40.6

0.7

HPV (Single andmultiple)Multiple HPV

HIV

*Appropriately adjusted based on comparison models

Conclusions

• Increased multiple infections in HIV+ women

• HR-HR and HR-LR-HR combinations increase risk of abnormalities compared to single

• Substantial proportion of risk reduced by removal of multiple HPV infections

Discussion

• Reasons for increased risk?

1. Do multiple HPV types infect same cell??-Enhanced oncogene products- increased transformation

2. Does risk change by combinations of oncogenic categories-biologic interactions- enhanced immunogenicity??

3. Any particular genotype combinations??

Discussion

• Cervical cancer-AIDS defining illness- proportion of risk potentially decreased-0.7%??????- Selection bias- majority of HIV- from colposcopy clinics

• Are HIV+ women subject to survival bias?- survivors cope with infections better

• Screening bias- convenience sample-underestimates or overestimates

Other epidemiologic issues

• Selection bias- Risk match or do not risk match HIV- women

• If we do match, can we make claims regarding genotypic prevalences?

• Information bias: are HPV risk categories correct, if not- non-differential misclassification

• Using cytology vs. histology- Non-differential misclassification

Future prospects

• Will HPV vaccines work??

Future plans

Graduate!!!!!

Dr.Hagensee and Dr.Kissinger (Mentors), Dr.Myer’sDr.Hagensee and Dr.Kissinger (Mentors), Dr.Myer’s

Hagensee Laboratory : Basic Hagensee Laboratory : Basic

Isaac SnowhiteIsaac Snowhite Joeli BrinkmanJoeli Brinkman Jennifer CameronJennifer Cameron

Melanie Palmisano Melanie Palmisano Anil ChaturvediAnil Chaturvedi Paula InserraPaula Inserra

Ansley HammonsAnsley Hammons Timothy SpencerTimothy Spencer

Clinical: Clinical:

Tracy BeckelTracy Beckel Liisa OakesLiisa Oakes Janine HalamaJanine Halama

Karen LenzcykKaren Lenzcyk Katherine LohmanKatherine Lohman Rachel HanischRachel Hanisch

Andreas TietzAndreas Tietz

LSUHSC:LSUHSC:

David Martin David Martin Kathleen DunlapKathleen Dunlap Patricia BralyPatricia Braly

Meg O’BrienMeg O’Brien Rebecca Clark Rebecca Clark Jeanne DumestreJeanne Dumestre

Paul FidelPaul Fidel

Acknowledgements