February 19, 2012

Research Roundup: Not all "high risk" HPV is high risk, but some strains are really fucking scary: a one-stop shop for strain-specific studies available on the Web.

Most of the the time when you get a test for "high risk HPV," what you get is a grab-bag test that tells you you're positive for one or more of 13 or 14 types of HPV, but doesn't tell you which type or types you're positive for. This is kinda informative, but not really that informative, because cervical cancer risk varies a lot within the so-called high-risk types. 

How much does it vary? Some studies (summary and links below) show a 40% chance of having CIN3 within two years if you have HPV-16 and an ASCUS Pap. But with HPV-56 and an ASCUS Pap, the chance of developing CIN3 within 2 years is only 1.9%. Either HPV-16 or HPV-56 will give you a positive result on a "high risk HPV" test, but your level of risk is about 20 times higher if you have HPV-16. 

HPV types also vary quite a bit in median clearance time and in their potential to cause different kinds of cervical cancer. 

General practitioners and even gynecologists don't always know much about type-specific risk. (They've got other things to do, like seeing umpteen patients a day and fighting with insurance companies.) When I got HPV genotyping (which I didn't specifically ask for), I found myself in a bizarre situation where my doctor was telling me I had HPV-31, 56, 66, and 83, sounded kinda confused and worried on the phone, and couldn't answer key questions like "Are these strains really bad?" and "Is it bad that I have multiple strains?" 

This post should help answer the first question. The bottom line is HPV-16, 18, 31, 33, 35, 45, 52, and 58 are significantly more worrisome than the other "high risk" types, and you should be especially vigilant on your own behalf if you have HPV-16, 18, or 45, because they tend to cause disease of the glandular cells inside the cervical canal, which is harder to detect on a Pap.  

(The answer to the second question is "probably not"--there doesn't seem to be any link between clearance time or progression to cervical cancer and infection with multiple strains.)

Type-specific HPV geekery after the jump, and my apologies if this is too much of a data dump...

...And as usual, some vocabulary and background first. 

Vocabulary
  • "HPV type" or "HPV strain" - a particular type of HPV, detectable by DNA testing. "HPV type" is slightly more correct, but both terms are in use and I'll use them interchangeably because I'm lazy.
  • "Clade" - a family of genetically-related HPV types. HPV types within the same clade sometimes have similar risk profiles or mechanisms for causing disease, and they sometimes have immunological cross-reactivity such that antibodies to one type in the clade will offer some protection against others. But not always. 
  • "Oncogenic" - Able to cause cancer. Often used interchangeably with "carcinogenic."
  • "High risk" - Able to cause CIN2 or CIN3. Some high-risk strains seem to "stall out" at CIN2 or CIN3, and are rarely associated with cervical cancer. So those strains would be classified by some researchers as "high risk" but not as "oncogenic." Not all researchers make this distinction. Research on the oncogenic potential of different high-risk strains is ongoing. 
Key background about HPV types
  • HPV is not one virus. Remember, HPV, and even high-risk HPV, is not one virus. There are 40-odd types of HPV that can infect the human genital tract. About half of those show some ability to cause cervical dysplasia and, if not treated, cervical cancer. And as I discussed in the vaccination post, exposure to one type does not give a person immunity to other types--and may not even give future immunity to that type. 
  • Persistence and oncogenic potential are not separate concepts. Studies often discuss the risk that particular types of HPV will progress to CIN2 or CIN3. Other studies, or sometimes the same studies, will often discuss the tendency of different HPV types to persist or to clear. These two concepts are not completely analytically distinct. HPV causes cervical cancer and precancerous changes by persisting. And it persists by integrating into the cervical cells in ways that can eventually cause cancer or precancerous changes. So HPV types with high persistence tend to be the same types with high rates of progression to CIN2 or CIN3. 
  • HPV Clade Groupings - Here is the breakdown of which HPV types are in which clade. Not all of the HPV types listed here are considered "high risk," but each clade listed contains at least one HPV type that has at some point been categorized as "high risk." The 14 types currently considered "high risk" on the Cervista or Hybrid Capture II HPV tests are bolded.  
  • For many types, the risk level is not yet clear. An example would be HPV-83. Some researchers categorize it as "high risk," but it's in the A-3 clade (not generally a high-risk grouping), and the type-specific research doesn't show it causing much CIN2 or CIN3. For a fuller discussion of possibly-oncogenic HPV types, see Classification of weakly carcinogenic human papillomavirus types: addressing the limits of epidemiology at the borderline, Schiffman et al., Infectious Agents and Cancer 2009, 4:8.
Population prevalence of HPV types
  • There are a ton of studies of prevalence of various HPV types among various populations. Often, studies rely on populations that will likely have higher HPV rates than the general population, for example, patients at an STD clinic, or patients who already have ASCUS or LSIL Paps. 
  • In addition, studies done since HPV vaccines were approved in 2006 may be expected to show lower rates of HPV-16 and HPV-18 infection because those types are covered by the available HPV vaccines. 
  • Probably the best pre-vaccine study of HPV type distribution in women in the US is Prevalence of HPV Infection Among Females in the United States, Dunne et al., JAMA, vol. 297, 2007. It's based on research conducted in 2003-2004 on a sample of about 1600 women from the general population. I say it's "probably the best" because it has a relatively large sample size drawn from the general population. This study found the following type-specific HPV infection rates. HPV types included in the grab-bag tests are, again, bolded
    • HPV-53 - 2.8%
    • HPV-52 - 2.3%
    • HPV-59 - 1.9%
    • HPV-66 - 1.6%
    • HPV-51 - 1.6%
    • HPV-16 - 1.5%
    • HPV-39 - 1.2%
    • HPV-56 - 1.0%
    • HPV-58 - 1.0%
    • HPV-18 - 0.8%
    • HPV-73 - 0.8%
    • HPV-31 - 0.7%
  • Note that this study did not report prevalence rates for high-risk HPV-33, -35, -45, and -68 because the prevalence rates were statistically unreliable, probably because there were lower numbers of women with those infections. In a Danish study of type-specific infection rates in women with normal Pap smears, HPV-33, -45, and -68 were about as prevalent as HPV-39, -56, and -51, giving a rough population prevalence of 1.0 - 1.4%. In that same study, HPV-35 was about half as prevalent as HPV-58 or -18, giving a rough population prevalence of about 0.5%. Note that what I'm doing here as I attempt to fill in this gap is total silly-wild-ass-guess apples-and-oranges mixing-and-matching of two different studies on two different populations. The Danish study, which I'll discuss more below, can be found in full text here:  http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2950170/  
OK, I think that's enough background (What?? No such thing! There's never enough background!). Here's what the current research is saying about type-specific HPV risk and clearance. 

Type-Specific Risk: Recent Studies

Study: Human Papillomavirus Genotypes and the Cumulative 2-Year Risk of Cervical Precancer, Wheeler et al., Journal of Infectious Diseases, 2006; 194:1291-9. 
Full text available? Yes, linked above. 
Starting point: ASCUS or LSIL Pap
Number of participants: 5060 (!!!) 
Duration: 2 years 

Summary: This study is based on the data collected during the Atypical Cells of Undetermined Significance/Low Grade Squamous Intraepithelial Lesions Triage Study (aka ALTS). ALTS is why today, if you have an ASCUS Pap, you get an HPV test and get referred for a colposcopy if the HPV test is positive. The primary goal of ALTS was to figure out whether HPV testing could be used to triage women with ASCUS, but it collected a shit ton of other useful data, including the data analyzed here. 

All study participants already had a mildly abnormal Pap smear (ASCUS or LSIL) at study entry. This study measured how many of them had CIN2 or CIN3 within 2 years, and broke that down by HPV type. Since the starting point was ASCUS/LSIL this study does NOT address the progression risk of HPV infection that has not yet caused an abnormal Pap. The study also ended after 2 years, so it likely failed to capture disease progression that happened after that. 

So here's what this study found: 


% women with CIN in 2 years
HPV Type
N
≥ CIN2
≥ CIN3
81
5
20.0
40.0
16
235
50.6
39.1
26
5
40.0
20.0
31
88
29.5
14.8
33
43
23.3
14.0
58
67
26.9
13.4
67
15
20.0
13.3
82
18
27.8
11.1
45
41
24.4
9.8
39
47
17.0
8.5
42
37
13.5
8.1
35
64
23.4
7.8
68
26
7.7
7.7
52
124
18.5
7.3
18
82
18.3
6.1
51
89
16.9
5.6
66
52
13.5
3.8
84
27
7.4
3.7
61
28
10.7
3.6
53
59
8.5
3.4
54
30
3.3
3.3
56
54
5.6
1.9
59
43
4.7
0.0
70
25
4.0
0.0
(What I've done here is bolded the 14 types of HPV covered by the grab-bag tests and re-ordered study Table 1 so that it lists HPV types in order from greatest percentage chance of ≥ CIN 3 within 2 years.)

A few things to notice here: 
  • The study authors calculated percentage of progression to CIN2 or worse and CIN3 or worse only for women with an infection with a single strain. So there's pretty good certainty here that the CIN was caused by the listed strain. 
  • Partly because multiple infections were excluded, the number of women infected with each type can be pretty small. (That's the "N" column.) Where N is small, the statistical validity of the conclusion isn't great. So keep that in mind. 
  • In this study, a lot of types that aren't usually classified as "high risk" are causing CIN2 and CIN3 at rates higher than some of the high risk strains. But those also tend to be the strains with a small number of infected women, so again, the statistical validity isn't great there.
  • Notice that HPV-16 is both more common than any other high-risk strain (235 women with a single-strain infection), and more dangerous than almost any other strain. That's how it causes so much cervical cancer. That's why it was one of the first strains to be targeted by vaccines. And that's why if you're not currently testing positive for HPV-16, you should seriously consider getting vaccinated. 

Study: Long-term Absolute Risk of Cervical Intraepithelial Neoplasia Grade 3 or Worse Following Human Papillomavirus Infection: Role of Persistence, Kjaer et al., Journal of the National Cancer Institute 2010; 102:1478-88. 
Full text available? Yes, linked above. 
Starting point: Normal Pap smear. 
Number of participants: 1222
Duration: up to 13.4 years

Summary: This study is a beast, and has a ton of tables and graphs that present the data in various ways. It also breaks down various HPV types by tendency to persist for 2 years, and then tendency to progress to CIN 2 or 3 after persisting. It was done in Denmark. Denmark has a national, Pathology Data Bank that retains information about cervical cytology and pathology. This study is different from the study above in a couple of major ways. First, it looked at women with normal Pap smears at study baseline. Second, it followed women for 13.4 years, rather than two. 

The bottom line looks like this:

HPV Type
% with ≥ CIN2 after 13.4 years
% with ≥ CIN3 after 13.4 years
16
28.5
26.0
18
15.4
15.4
33
19.1
12.8
31
15.7
9.8
35
18.2
9.1
58
16.7
8.3
51
8.6
6.9
45
8.5
6.4
52
4.7
4.7
56
2.8
2.3
39
3.6
0
59
0
0
68
0
0
53
0
0
66
0
0


(What I've done here: I combined the data from study Table 2 and Table 3, pulling out only data for single-type HPV infections (to make this data parallel to the ALTS data above), I sorted it by greatest percentage of CIN 3, and I bolded the HPV types covered by commonly-available HPV tests.) 


StudyHuman papillomavirus type distribution in 30,848 invasive cervical cancers worldwide: Variation by geographical region, histological type and year of publication, Li et al., International Journal of Cancer, Volume 128, Issue 4, pages 927–935, 15 February 2011.
Full text available? Yes, linked above.
Starting point: Invasive cervical cancer
Number of participants: 30,848 

Summary: The two studies above looked at women who had HPV and tracked whether they developed CIN 2 or 3. This study pooled data on women who had invasive cervical cancer and looked at what HPV types were causing that cancer. The discussion is well worth reading for the breakdown of which types of HPV cause cancer in which areas of the world--it varies a bit, and it's good background to keep in mind as you look at other studies done in different locations. 

The key table to look at is Table 2, which breaks out which types of HPV were present in squamous cell cancer (SCC) and adenocarcinoma (ADC). (Summary below.)

HPV Type
% SCC caused
% ADC caused
16
59.3
36.3
18
13.2
36.8
58
5.1
1.5
33
4.9
2.2
45
4.4
5.2
31
4.0
2.3
52
3.6
1.2
35
1.9
0.6
59
1.4
0.8
39
1.3
0.8
51
1.0
0.6
56
0.8
0.2
68
0.5
0.5
66
0.4
0.1

This study tells you why HPV-16 and 18 were targeted for the first round of vaccine development: eliminate them, and you eliminate 72.5% of squamous cell cancer and 73.1% of adenocarcinoma. 

Type-Specific Persistence and Clearance Rates: Recent Studies

Study: Persistence of Genital Human Papillomavirus Infection in a Long-Term Follow-Up Study of Female University Students, Sycuro et al., The Journal of Infectious Diseases 2008: 198. 
Full text available? Yes, linked above. 
Starting point: HPV infection. 
Number of participants: 147 
Duration: 4-12 years

Summary: This small study looked at women who had first tested positive for HPV while they were undergraduates, followed them for 4-12 years, and looked at factors for persistence of a single-type HPV infection over that time. Study Table 3 is where the type-specific information is summarized, and it looks like this: 

HPV Type
N
% Long-term persistent
16
59
8.5
18
23
4.4
31
20
10.0
33
12
0.0
35
4
0.0
39
31
6.5
45
9
0.0
51
30
0.0
52
26
15.4
53
26
3.9
56
35
0
58
12
0
59
20
0
66
24
16.7
68
3
0

Note that the small number of participants is a major weakness of this study. 


Study: High-risk HPV type-specific clearance rates in cervical screening, Bulkmans et al., British Journal of Cancer (2007) 96, 1419-1424. 
Full text available? Yes, linked above
Starting point: Normal Pap or "Borderline or Mild Dyskaryosis" (BMD), the UK equivalent of ASCUS/LSIL Pap.
Number of participants: about 1500  
Duration: 18 months

Summary: This study looked at type-specific HPV clearance rates in women with normal Pap smears and women with "BMD" Paps, roughly the UK equivalent of an ASCUS or LSIL Pap. Table 1 and 2 in the study are where the action is. The take-home lesson is that HPV-16, 18, 31, and 33 are very slow to clear (particularly 16 and 31), and that infections are slower to clear in women with BMD smears at baseline. Table 3 of the study also summarizes progression rates to CIN3 in women who did not clear their HPV type.

StudyA 2-Year Prospective Study of Human Papillomavirus Persistence among Women with a Cytological Diagnosis of Atypical Squamous Cells of Undetermined Significance or Low-Grade Squamous Intraepithelial Lesion, Schiffman et al, Journal of Infectious Diseases (2007) 195 (11): 1582-1589. 
Full text available? Yes, linked above. 
Starting point: ASCUS or LSIL Pap
Number of participants: 4500  
Duration: 2 years

Summary: This is another study that crunches data from ALTS. Figure 2 is where the action is:


Figure 2.


What you see is a similar clearance curve for each type, but a 24-month clearance rate that varies quite a bit by type, from what looks like over 95% clearance for HPV-51, to something around 80% clearance for HPV-52. This is another study worth digging into. 

StudyHuman papillomavirus (HPV) types 16, 18, 31, 45 DNA loads and HPV-16 integration in persistent and transient infections in young women, Ramanakumar et al., BMC Infectious Diseases 2010; 10: 326.
Full text available? Yes, linked above. 
Starting point: HPV infection
Number of participants: 636  
Duration: 2 years

Summary: This study focused only on HPV-16, 18, 31, and 45. It found that once detected, 30% of HPV-16, 44% of HPV-18, 63% of HPV-31 and 55% of HPV-45 infections cleared within 6 months.

Type-Specific Risk: Adenocarcinoma

HPV can cause two different types of cervical cancer: squamous cell carcinoma, and adenocarcinoma. Squamous cell carcinoma affects the sqamous cells that line the exterior of the cervix and may extend partway into the cervical canal. Adenocarcinoma affects the glandular cells that line the cervical canal. 

HPV types in clade A-7, particularly HPV-18 and HPV-45, are more strongly oncogenic for adenocarcinoma than strains from other clades. However, HPV-16 causes roughly as much adenocarcinoma as HPV-18 does. (See Human papillomavirus type distribution in 30,848 invasive cervical cancers worldwide: Variation by geographical region, histological type and year of publication, Li et al., International Journal of Cancer, Volume 128, Issue 4, pages 927–935, 15 February 2011.) 

HPV-16, 18, 45, 59, and 35 cause approximately 96% of all cervical adenocarcinoma. (See Worldwide Human Papillomavirus Etiology of Cervical Adenocarcinoma and Its Cofactors: Implications for Screening and Prevention, Castellsague et al., Journal of the National Cancer Institute, Vol. 98, No. 5, 2006.)

Adenocarcinoma is particularly troubling for a few reasons:
  • Because adenocarcinoma is located higher in the cervical canal, it is more difficult to detect adenocarcinoma by Pap screening. 
  • This means adenocarcinoma tends to be detected later, at a stage where treatment is more difficult and more invasive. (See http://www.ncbi.nlm.nih.gov/pubmed/16175089)
  • The absolute rate of adenocarcinoma has been rising. 
What this means to you: If you have a positive test for one of the HPV types most likely to cause adenocarcinoma (HPV-16, 18, and 45 particularly), you may want to be a more vigilant, assertive patient about the whole thing. For example:
  • A lot of times if your Paps start coming back normal, your doctor will stop doing HPV tests--even without ever getting a negative HPV test. If you've had HPV-16, 18, or 45, you may want to get a little bitchy and insist on another HPV test to make sure the HPV has really cleared and isn't causing hard-to-detect problems in the cervical canal. 
  • And then if that HPV test comes back positive, and especially if it keeps coming back positive year after year, you may want to insist on a colposcopy with an endocervical biopsy every year or two, just to make sure the cells deep in the cervical canal actually get sampled periodically. 
For women over 30 with a normal Pap but persistent infection with HPV-16 or -18, this is actually what the American Society for Colposcopy and Cervical Pathology (ASCCP) recommends, but not all doctors follow these recommendations.  

Not enough? Want more studies?

If you want to read a more comprehensive range of studies on HPV types or on a specific type, you can do two things:

1) Mine the bibliographies of the studies discussed above. If you go to the html full text version, most of the bibliographies (listed at the end of the study) will have links to the PubMed abstract of the other studies, and sometimes links to free full-text versions. Other free full-text versions may be linked from the PubMed abstract page. (Some studies aren't available to the public in full text because they've been sold to greedhead for-profit journals who charge academic libraries something like $25,000 a year for a subscription. A subscription allowing the school's students and staff to access studies done by the school's own researchers.)

2) Do your own search on PubMed (linked from my blogroll). If you're looking for studies discussing a particular HPV type, do a search for, "HPV 45" or "HPV 33" in quotes like that, plus the word "cervix". You'll get way too many results, but you can winnow them down eventually. If you leave out "cervix" as a search term, you'll get results about other kinds of disease caused by these HPV types. Those can be interesting too. 

No comments: