The Clinical Effectiveness and Cost-effectiveness of Screening for Anal Squamous Intraepithelial Lesions in Homosexual and Bisexual HIV-Positive Men FREE

The incidence of anal squamous cell carcinoma (SCC) in homosexual and bisexual men exceeds that in heterosexual men^1- 5 and is higher than that of cervical cancer in women.⁶ Estimates of the annual anal SCC incidence among homosexual men before the acquired immunodeficiency syndrome (AIDS) epidemic were between 12.5 and 36.5 per 100,000,³ and the incidence may be even higher in homosexual and bisexual men infected with human immunodeficiency virus (HIV).^7- 8

The epithelium in both the cervix and anus may contain atypical squamous cells of undetermined significance, low-grade squamous intraepithelial lesions (SIL), and high-grade SIL.⁹ In the cervix, it is fairly well established that high-grade SIL represent the identifiable precursor to invasive cervical SCC.¹⁰ Given the similarities between cervical and anal SIL and SCC, including a strong association with anogenital human papillomavirus (HPV) infection, high-grade anal SIL likely represent the precursor to anal SCC.

Homosexual and bisexual men with HIV are at increased risk for persistent HPV infection and HPV-associated anal squamous intraepithelial lesions (ASIL), with prevalence rates of ASIL ranging from 20% to 45%.^8,11- 17 Risk factors for ASIL include a lower CD4 cell count, reflecting more advanced immunosuppression, and HPV infection.^12,18- 19 Recent studies of the natural history of anal disease have shown that ASIL may progress to a higher grade lesion in a short time, and that regression of high-grade ASIL is rare.^15- 16

These data, in conjunction with recent reports of increased anal SCC among men with AIDS,^7- 8 suggest that HIV-positive homosexual and bisexual men are at significant risk for developing anal SCC. Furthermore, unlike most malignancies in HIV-positive men, anal SCC may be preventable. Based on the model of cervical SCC screening, anal cytology has been suggested as a potential screening test to prevent the development of anal cancer in these high-risk men.^20- 22 Our objective was to estimate the clinical benefits (life expectancy and quality-adjusted life expectancy), costs (dollars), and cost-effectiveness of screening HIV-positive homosexual and bisexual men for ASIL and SCC.

We developed a state-transition Markov model to calculate lifetime costs, life expectancy, and quality-adjusted life expectancy for no screening (the status quo) and the following screening strategies using anal Papanicolaou (Pap) tests: every 3 years, every 2 years, every year, and every 6 months. We evaluated lifetime screening strategies for men presenting with (1) CD4 cell counts higher than 0.50×10⁹/L (500/µL), (2) CD4 cell counts of 0.20 to 0.50×10⁹/L, and (3) CD4 cell counts below 0.20×10⁹/L. Performance of alternate screening strategies was compared using the incremental cost-effectiveness ratio, defined as the additional cost of a specific screening strategy, divided by its additional clinical benefit, compared with the next least-expensive strategy. Screening strategies that were less effective and more costly than an alternate strategy (ie, strongly dominated) and strategies with a higher incremental cost-effectiveness ratio than a more effective alternative strategy (ie, weakly dominated) were ruled out.²³ The analysis was performed from the societal perspective and followed the reference case recommendations of the Panel on Cost-Effectiveness in Health and Medicine.²³ Morbidity and mortality consequences were captured in a single measure using quality-adjusted life years (QALYs). Future costs and QALYs saved were discounted at an annual rate of 3%. Sensitivity analyses were performed to determine the robustness of the cost-effectiveness results in the face of reasonable variation in the underlying data assumptions.

Markov models, such as this one, depict the natural history of disease as an evolving sequence of mutually exclusive health states, defined to capture important clinical characteristics. Each health state describes the natural history of HIV disease (CD4 cell counts >0.50×10⁹/L, 0.20-0.50×10⁹/L, and <0.20×10⁹/L), natural history of anal disease (normal, atypical squamous cells of undetermined significance, low-grade ASIL, high-grade ASIL, anal SCC), and screening status (screened positive or negative for atypical squamous cells of undetermined significance, low-grade or high-grade ASIL, anal SCC, and not screened). Men start in the model as unscreened and are initially distributed into health states according to cross-sectional data on ASIL prevalence.²⁴ The time horizon of the analysis is divided into equal increments, referred to as Markov cycles, during which men transition from one state to another. The model then uses transition probabilities, derived from a review of the literature, to move men through different health states over time. For example, in any given month men may have progression or regression of anal disease, HIV disease, or both. The probability of ASIL progression is conditional on the stage of HIV disease, with men in the earliest stages at the lowest risk of progression. Each month men may die of an HIV-related illness, anal SCC, or other causes.

In a designated screening month, an abnormal screening test result (defined as atypical squamous cells of undetermined significance, low-grade or high-grade ASIL) triggers a diagnostic anoscopy and biopsy. An individual with low-grade ASIL will be followed up with surveillance screening every 6 months, while an individual with high-grade ASIL will undergo surgical treatment. Men with false-positive results accrue the costs of a diagnostic workup (eg, anoscopy and biopsy) before resuming a regular screening schedule. In the absence of screening, men who develop invasive anal SCC have an annual probability of symptoms that will cause them to seek diagnostic testing and treatment.

Table 1 summarizes selected parameter estimates.^{3- 4,7- 8,11- 13,15- 19,22,24- 45} The base case estimates of ASIL incidence, progression, and regression were derived from published data from the 2 largest prospective cohort studies, both with well-defined control groups, in San Francisco, Calif,^15- 16,19 and Seattle, Wash.^12- 13 Since there are no studies from which to directly derive the probability of high-grade ASIL progression to cancer, we constructed Markov models to simulate the natural history of ASIL in HIV-negative homosexual and bisexual men,¹⁶ and derived the progression rate of high-grade ASIL that would be required to match cancer incidence rates reported by Daling and coworkers.³ We adjusted this estimate using the relative risk of anal SCC in HIV-positive vs HIV-negative men.^7- 8 We assumed 30% of men with invasive anal cancer would develop symptoms causing detection of disease each year.^38- 39 Standard treatment of anal cancer is a protocol of combined chemotherapy and percutaneous radiotherapy.⁴⁶ We assumed that HIV-positive men with anal SCC experience the same efficacy and toxic effects with cancer treatment as HIV-negative men with anal SCC.^47- 50

We made several conservative assumptions that biased the analysis against screening: (1) incidence and progression of ASIL in men with CD4 cell counts below 0.20×10⁹/L were no higher than in those with CD4 cell counts between 0.20 and 0.50×10⁹/L; (2) regression of low-grade ASIL exceeded 50% over 2 years, similar to regression rates reported in women with cervical SIL⁵¹; (3) men treated for high-grade ASIL were not cured but had a 75% lower monthly risk of progression to invasive anal SCC after treatment; (4) a temporary quality of life of 0.0 (equivalent to death) was assigned to the 7 days following treatment for high-grade ASIL, reflecting the postoperative pain reported by patients; and (5) men who developed anal SCC and were detected through screening were not diagnosed at earlier stages of cancer (ie, no stage shift resulted from screening).

Estimates of HIV progression were based on data reflecting less intensive use of antiretroviral therapy than is now standard.^33- 35,37 We used more recent data to project the impact of highly active antiretroviral therapy on the results, incorporating slower HIV progression and decreased HIV-related morbidity and mortality.^52- 54 Since the effect of highly active antiretroviral therapy on both HPV-mediated anal disease and long-term survival is uncertain, we conducted this analysis as a sensitivity analysis.^55- 56

Estimates of the diagnostic performance of anal cytology have been variable,^57- 60 with inconsistent use of anoscopy for histological confirmation.^27,60- 61 We used data from the largest prospective study of ASIL in homosexual and bisexual HIV-positive men,²² which reported an anal Pap test sensitivity and specificity of 81% and 63%, respectively.

No published studies have used preference-weighted scales to assess quality of life in homosexual and bisexual HIV-positive men with anal disease. We used quality weights derived by Freedberg and coworkers³⁶ for health states characterized by CD4 cell counts higher than 0.20×10⁹/L (0.94) and lower than 0.20×10⁹/L (0.84). Although these weights were not elicited using the preferred methods of the standard gamble or time trade-off, they were stratified by CD4 cell count and included a health state with AIDS and AIDS-related cancer (0.56). We also applied utilities reported by Tsevat et al⁶² using the time trade-off method (from 0.79 to 0.87 depending on stage of HIV disease) in sensitivity analysis.

The costs of anal cancer screening, diagnosis, and treatment were estimated using Medicare average allowed charges as a proxy for true resource costs.^42,63 The resource use associated with ASIL treatment was based on clinical care algorithms developed by the University of California at San Francisco Faculty Practice. Low-grade ASIL required a diagnostic anoscopy and biopsy of any abnormal lesion but no treatment. High-grade ASIL required referral to a surgeon, 1 preoperative office visit with anoscopy, electrocautery, or excisional biopsy through a colposcope with anesthesia in an outpatient surgical setting, postoperative oral analgesia for 2 weeks, 1 postoperative visit at 30 days, and subsequent follow-up every 6 months with anoscopy.

The costs of initial treatment, continuing care, and terminal care for anal cancer were approximated based on published costs for colorectal cancer.⁴³ We also used alternative sources to estimate costs based on a discharge diagnosis of anal cancer from the Healthcare Cost and Utilization Project,⁴⁴ which were similar to the published estimates for colon cancer. We approximated average patient time costs incorporating travel, waiting time, and direct care: screening (65 minutes), low-grade ASIL diagnostic workup (2 hours), high-grade ASIL workup, treatment, and postoperative recovery (40 hours), and invasive anal SCC (120 to 160 hours).^23,45,64 Since wage rates for this target population are uncertain, we did not include time costs in the base case but incorporated these in sensitivity analysis.

Costs of monthly HIV care were based on estimates developed by Freedberg and coworkers from the AIDS Cost and Services Utilization Survey.^{32,36,65- 66} In our sensitivity analysis of men receiving highly active antiretroviral therapy, we modified these costs to include zidovudine, indinavir, lamivudine, and quarterly viral load testing.^53,67 Costs were updated to 1997 US dollars using the medical care component of the Consumer Price Index from the Bureau of Labor Statistics.⁶⁸

The projected clinical and economic outcomes of individual screening strategies are shown in Table 2. If screening was initiated early in HIV disease (CD4 cell count >0.50×10⁹/L), an anal Pap screening every 2 years increased quality-adjusted life expectancy by 2.7 months and increased total costs by $2940, resulting in an incremental cost-effectiveness ratio of $13,000 per QALY saved compared with no screening. Screening annually with anal Pap tests provided additional clinical benefit for $16,600 per QALY saved compared with screening every 2 years. Increasing screening frequency to twice annually provided less than 3 days of additional quality-adjusted life expectancy compared with annual screening, with a cost-effectiveness ratio of $49,600 per QALY saved.

If screening was not initiated until later in HIV disease (CD4 cell count, 0.20-0.50×10⁹/L), anal Pap test screening yearly dominated screening every 2 years, with a cost-effectiveness ratio of less than $25,000 per QALY saved compared with no screening. Increasing screening frequency to twice yearly provided minimal additional benefit compared with annual screening. Results unadjusted for health-related quality of life were similar.

Estimates of cost-effectiveness were most influenced by the rate of progression of high-grade ASIL to invasive anal SCC and the efficacy of treatment for high-grade ASIL (Figure 1 and Figure 2). Results were less sensitive to incidence, progression, and regression of low-grade ASIL; Pap test sensitivity and specificity; quality of life for cancer and HIV; compliance with screening; costs of screening tests, patient time, and medical care for cancer and HIV disease; and the discount rate. Selected univariate sensitivity analyses are shown in Table 3.

Despite varying incidence, progression, and regression of ASIL±50% from the base case, the cost-effectiveness ratio of screening every 2 years remained between $12,000 and $14,000 per QALY, and of yearly screening between $15,000 and $20,000 per QALY (Table 3). As progression rates of high-grade ASIL were increased relative to the base case, the cost-effectiveness of all screening strategies became more attractive (Figure 1). The converse was true as progression rates were lowered, although even with a monthly progression rate of high-grade ASIL to anal SCC as low as 1 per 1000, the cost-effectiveness ratio for screening every 3 years was less than $30,000 per QALY.

In the base case, we assumed treatment for high-grade ASIL would be 75% effective (ie, 75% reduction in progression to cancer). If treatment efficacy was higher (similar to that observed for high-grade cervical SIL in women), the cost-effectiveness of yearly screening fell below $10,000 per QALY (Figure 2). However, even if treatment of high-grade ASIL were only 25% effective, the cost-effectiveness ratios for every 3-year and 2-year screening remained less than $40,000 per QALY.

In the absence of highly active antiretroviral therapy, if screening was not initiated until CD4 cell counts were below 0.20×10⁹/L, the lifetime clinical benefits (0.3 quality-adjusted months) were smaller than when screening was initiated earlier, but still exceeded the typical life expectancy gains with prophylaxis for Mycobacterium avium complex (0.18 quality-adjusted months) (Table 2).³⁶ To project the impact of highly active antiretroviral therapy in late HIV disease (CD4 cell count <0.20×10⁹/L), we conducted a sensitivity analysis using data from the AIDS Clinical Trials Group Protocol 320.⁵² We explored the implications of viral suppression lasting 1, 2, or 3 years and found that annual Pap screening had an incremental cost-effectiveness ratio of $21,900, $21,700, or $20,600 per QALY saved, respectively, compared with $57,000 per QALY saved in the absence of highly active antiretroviral therapy.

We also used data reported by Gulick et al⁵³ to estimate the impact of highly active antiretroviral therapy started earlier in HIV disease, when the CD4 cell count was between 0.20 and 0.50×10⁹/L. Assuming the duration of viral load suppression was 2 years, the cost-effectiveness of annual Pap screening was $17,300 per QALY saved. If viral load suppression was sustained for longer periods, screening became even more cost-effective. If the use of highly active antiretroviral therapy, in addition to decreasing HIV-related morbidity and mortality, lowered the risk of incident high-grade ASIL and/or the transition rate of high-grade ASIL to cancer, these ratios remained attractive. For example, the cost-effectiveness ratio of annual screening remained below $50,000 per QALY even if progression of high-grade ASIL to invasive cancer was reduced to 25% of the base case.

There is substantial evidence that HIV-positive homosexual and bisexual men are at increased risk for ASIL, persistent HPV infection, and invasive anal SCC.^8,11- 17 Our results indicate that screening homosexual and bisexual HIV-positive men for ASIL, at all stages of HIV, would prolong quality-adjusted life expectancy. If screening was initiated early in HIV disease (CD4 cell count >0.50×10⁹/L), the cost-effectiveness ratio for anal Pap screening every 2 years was robust over a wide range of sensitivity analyses. If screening was initiated in later HIV disease (CD4 cell count <0.50×10⁹/L), then annual screening was preferred due to the greater amount of prevalent anal disease. Screening every 6 months provided little additional benefit over that of annual screening in nearly all sensitivity analyses.

The most critical factors influencing the cost-effectiveness of screening were the progression rate of high-grade ASIL to invasive anal SCC and the effectiveness of therapy for high-grade ASIL. Even so, at monthly progression rates as low as 1 per 1000 and treatment efficacy as low as 25%, screening with anal cytology provided life expectancy benefits for a reasonable cost. Furthermore, if there is a survival benefit associated with diagnosis at earlier stages of anal SCC,⁶⁹ the cost-effectiveness ratios will be even more attractive. In contrast, changes in the rates of incidence, progression, and regression of low-grade ASIL had little impact on our major results.

Cervical cancer screening is considered one of the great successes among cancer screening programs.⁵¹ There are important differences between existing screening practice in women for cervical SIL and proposed screening in HIV-positive homosexual and bisexual men for anal SIL. In women, low-grade cervical SIL are often treated despite the fact that a substantial proportion may spontaneously regress. In fact, the associated costs of colposcopic evaluation and interventional therapy for low-grade cervical SIL have been estimated at close to $6 billion annually.⁷⁰ This analysis assumes that men with low-grade ASIL, aside from careful follow-up, receive no intervention.

There are several limitations to this analysis. The natural history of ASIL is uncertain. However, even when regression rates were twice that of our base case and progression rates half of our base case, the cost-effectiveness results for biennial and annual screening remained attractive. Health-related quality-of-life measures using preference-weighted scales in homosexual and bisexual men are uncertain, although our sensitivity analysis showed little impact on quality of life on the major results. Finally, there are no published anal SCC costs and we relied on estimates for colorectal cancer as a proxy.⁴³ However, sensitivity analysis demonstrated that even if cancer costs were as low as 25% or as high as 200% that of our base case, the policy conclusions did not change.

In the setting of highly active antiretroviral therapy, suppression of viral replication can be sustained for at least 2 years⁵³ and in patients with CD4 cell counts below 0.20×10⁹/L progression to AIDS and death have been reduced.⁵² Although we found the cost-effectiveness of anal cancer screening to remain favorable when we incorporated these data, our estimates are preliminary since the long-term impact of potent combination antiretroviral drugs is not yet known. If we further consider the real-world failure rates of antiretroviral regimens due to resistance and nonadherence, the lifelong effect is even more uncertain.⁵⁶ The impact of highly active antiretroviral therapy on the incidence of new anal neoplasia and progression of established neoplasia is also uncertain. However, if the risk of ASIL is not substantially decreased with newer antiretroviral treatments, screening will be even more cost-effective because of the longer life expectancy of men with sustained viral suppression.

Several policy barriers exist to developing screening programs for ASIL and anal SCC. There is a shortage of trained clinicians to perform anoscopy, biopsy, and treatment of ASIL. Anal Pap test screening is not routinely performed and may also require training. Optimal diagnostic methods and treatment regimens for high-grade disease are uncertain. Therefore, educational efforts with respect to screening and follow-up will be needed for patients, clinicians, and policymakers.

In conclusion, using the best data currently available, over a broad range of parameter estimates and with a number of conservative assumptions to bias the analysis against screening, screening HIV-positive homosexual and bisexual men with anal cytology was associated with substantial clinical benefit. Regardless of when screening was initiated, the cost-effectiveness of either a yearly or every-2-year screening schedule was comparable with other accepted preventive measures in clinical medicine.

We recommend that immediate consideration be given to identifying the real-world barriers associated with implementing such a screening policy. In addition, priorities for future research include identifying rates of progression of high-grade ASIL to cancer with and without highly active antiretroviral therapy, evaluating treatment modalities for high-grade ASIL, and assessing acceptability of screening to both patients and health care providers.