Recent and projected incidence trends and risk of anal cancer among people with HIV in North America

Abstract

Background

Anal cancer risk is elevated among people with HIV. Recent anal cancer incidence patterns among people with HIV in the United States and Canada remain unclear. It is unknown how the incidence patterns may evolve.

Methods

Using data from the North American AIDS Cohort Collaboration on Research and Design, we investigated absolute anal cancer incidence and incidence trends nationally in the United States and Canada and in different US regions. We further estimated relative risk compared with people without HIV, relative risk among various subgroups, and projected future anal cancer burden among American people with HIV.

Results

Between 2001 and 2016 in the United States, age-standardized anal cancer incidence declined 2.2% per year (95% confidence interval = ‒4.4% to ‒0.1%), particularly in the Western region (‒3.8% per year, 95% confidence interval = ‒6.5% to ‒0.9%). In Canada, incidence remained stable. Considerable geographic variation in risk was observed by US regions (eg, more than 4-fold risk in the Midwest and Southeast compared with the Northeast among men who have sex with men who have HIV). Anal cancer risk increased with a decrease in nadir CD4 cell count and was elevated among those individuals with opportunistic illnesses. Anal cancer burden among American people with HIV is expected to decrease through 2035, but more than 70% of cases will continue to occur in men who have sex with men who have HIV and in people with AIDS.

Conclusion

Geographic variation in anal cancer risk and trends may reflect underlying differences in screening practices and HIV epidemic. Men who have sex with men who have HIV and people with prior AIDS diagnoses will continue to bear the highest anal cancer burden, highlighting the importance of precision prevention.

Anal cancer risk is nearly 20 times higher among people with HIV than in the general population of the United States (1). HIV contributes disproportionately (33%) to anal cancer burden among men; among women, 3% of cases are HIV attributable (2,3). Although anal cancer incidence among the general US population has notably increased (4-7), recent trends among people with HIV remain unclear. Furthermore, information about anal cancer incidence among people with HIV in Canada is limited, with only 1 administrative database study from Ontario (8).

Anal cancer incidence increases with age, and it remains highest among people with HIV who have prior AIDS diagnoses (9-11). Since the advent of antiretroviral therapy (ART), survival among people with HIV has improved, leading to a growing number of people with HIV beyond 50 years of age (12,13). The growing population of aging people with HIV, many of whom had prior AIDS diagnoses, portends a substantial shift in a typical profile of anal cancer patients with HIV (12,14). The anticipated anal cancer burden with HIV, however, particularly in the context of aging people with HIV and in those individuals with previous AIDS diagnoses, remains unclear.

Recent evidence from the Anal Cancer/HSIL Outcomes Research (ANCHOR) study indicating that anal precancer treatment reduces anal cancer risk among people with HIV and data on risk stratification and screening accuracy provided a critical foundation to generate the consensus-based guidelines for anal cancer screening (11,15-19). Epidemiological estimates of absolute risk, relative risk, and recent and projected trends in anal cancer incidence among people with HIV could inform future updates to screening guidelines and clinical care practice.

Using data from the largest North American HIV cohort study consortium, the North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD), we described recent trends in anal cancer incidence among people with HIV in the United States and Canada. Given the inconsistent use of anal cancer screening in the United States (confined mainly to the Western and Northeast regions) (4,20) and notable geographic differences in the underlying HIV epidemic, we further explored evidence of regional variation in anal cancer risk by estimating absolute and relative risk and calendar trends by geographic regions in the United States. Finally, using a validated mathematical model, we projected the anal cancer incidence patterns and burden among people with HIV from 2016 to 2035, according to age and prior AIDS diagnosis.

Methods

Study design and population

North American AIDS Cohort Collaboration on Research and Design

NA-ACCORD is a collaborative study that pools individual-level data from 22 HIV clinical and interval cohorts in Canada and the United States (21,22). It includes people with HIV 18 years of age and older with at least 2 clinic visits within the first year of cohort enrollment. Design, data collection, and data harmonization methods have previously been described (21,22). Each cohort reported demographic, medication, laboratory, diagnoses, and vital status data annually to the Data Management Core for data harmonization and quality control. Incident anal cancer diagnoses were ascertained separately by each cohort from medical records, patient interviews (confirmed by medical record reviews), or linkage with a cancer registry. We excluded 9.6% of participants because of missing data on CD4 cell count and viral load (Supplementary Figure 1, available online).

Mathematical model (HIV natural history framework)

We developed a microsimulation model to project the anal cancer incidence and burden among people with HIV in the United States from 2017 to 2035. The model (Supplementary Figure 2, available online) tracks each person (population characteristics described in Supplementary Methods 1 and Supplementary Tables 1-3, available online), starting from the point of HIV acquisition, and encompasses disease progression, diagnosis, initiation of treatment, treatment adherence, and both HIV-specific and other-cause mortality. Receiving HIV treatment and treatment adherance was associated with improvement in CD4 cell count (Supplementary Figure 3, available online). The model accurately captured the decreased occurrence of new HIV infections and increased likelihood of linkage to care over calendar years in the United States (Supplementary Figures 4 and 5, available online). For model calibration, we employed a heuristics-based multistep approach to estimate unobservable parameters and those parameters with high variability (Supplementary Tables 4-8, Supplementary Figure 6, available online). Model validation was performed by comparing model-predicted outcomes against the observed data on HIV prevalence, AIDS prevalence, and deaths among people with HIV (Supplementary Figures 7-9, available online).

Statistical analysis

Person-time at risk of developing anal cancer accrued from the latest of the first date of participants’ enrollment into NA-ACCORD, first available CD4 cell count or HIV RNA level measurement, or the beginning of cohort-specific cancer diagnosis ascertainment. Participants no longer contributed person-time at the first date of anal cancer diagnosis; death; December 31, 2016; or end of cohort-specific cancer ascertainment. We calculated anal cancer incidence rates determined as the number of cancer cases per 100 000 person-years in North America, in the United States (overall and by regions—Northeast, West, Midwest, and South), and in Canada. Incidence rates were age standardized using the 2010 population of people with HIV in the United States as the standard population. Incidence was estimated by age (<35, 35-44, 45-54, and ≥55 years), sex (male, female), HIV transmission group (men who have sex with men [MSM], people who inject drugs, heterosexual transmission, and other or unknown), race and ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, others), prior opportunistic illnesses (yes or no), number of opportunistic illnesses (0, 1, 2), nadir CD4 cell count (<100, 100-200, 200-349, and ≥350 cells/mm³), comorbidities (chronic kidney disease, diabetes, cardiovascular disease, hypertension), and number of comorbid conditions (0, 1, 2, ≥3).

We also estimated nonparametric cumulative anal cancer risk curves for MSM and non-MSM men, overall and stratified by the number of opportunistic illnesses and nadir CD4 cell count, considering death as a competing event. Pointwise confidence intervals were derived for cumulative incidence estimates at 5, 10, and 15 years of follow-up.

To evaluate time trends in anal cancer incidence, we estimated annual percentage changes (23). JoinPoint, version 4.8.0.1, software (National Cancer Institute, Bethesda, MD) was used to estimate piecewise–log-linear trends and derive annual percentage changes. P values were estimated using the permutation distribution of the test statistic (statistical significance at P < .05, 2-sided). We used Poisson regression to estimate incidence rate ratios to compare anal cancer risk among major subgroups (MSM and non-MSM men) in the United States. Multivariable models included US regions, attained age group, race and ethnicity, number of opportunistic illnesses, nadir CD4 cell count, comorbidities, and number of comorbid conditions. Finally, we estimated rate ratios compared with people without HIV (derived by removing estimated anal cancer cases with among people with HIV from anal cancer incidence among the general population) (details in Supplementary Methods 2, available online) (2).

Mathematical modeling

Anal cancer incidence projections were modeled using a first-order state-transition (microsimulation) model of a representative US HIV population (Supplementary Methods, available online). The model combined observed anal cancer incidence rates (for MSM, non-MSM men, and women, stratified by age and prior AIDS diagnosis) and overall calendar trends in incidence among people with HIV with the population size of people with HIV (from the HIV natural history framework) and considered competing mortality risks attributable to HIV/AIDS and other causes. AIDS diagnosis was defined as having 1 or more opportunistic illness or nadir CD4 cell count under 200 cells/mm³. Incidence-based transition probabilities were fitted by age, transmission group, and AIDS diagnosis status. The mathematical model was developed in R (R Foundation for Statistical Computing, Vienna, Austria).

Results

A total of 111 566 participants with 1 035 692 person-years of follow-up (median [inter quartile range] = 8.3 [3.8-13.9] years) from 17 cohorts were included. A total of 639 incident anal cancer cases were observed in NA-ACCORD. The majority (619 [97%]) were men, particularly (43%) MSM. There were 20 (3%) anal cancer cases among women with HIV and 100 (16%) cases among people who inject drugs (Supplementary Table 9, available online). Most cases had at least 1 prior opportunistic illness (321 [51%]). Among cases with prior opportunistic illnesses, 66% had 2 illnesses; 342 [54%] had at least 2 other comorbid conditions (chronic kidney disease, diabetes, cardiovascular disease, or hypertension).

Overall, during 2001-2016 in North America, age-standardized anal cancer incidence (per 100 000 person-years) among people with HIV was 61.4 (95% confidence interval [CI] = 56.4 to 66.6) (Table 1). Incidence was greater (n = 51; 99.2, 95% CI = 69.7 to 139.0) in Canada than in the United States (n = 584; 60.4, 95% CI = 55.4 to 65.8). Among MSM with HIV in Canada, incidence was 154.1 (n = 44; 95% CI = 103.7 to 223.1); in the United States, it was 85.2 (n = 229; 95% CI = 72.4 to 99.8) (Supplementary Table 10, available online). In the United States, large regional variation in incidence was observed, ranging from 29.3 (95% CI = 21.8 to 38.7) in the Northeast to 89.7 (95% CI = 71.6 to 110.9) in the Midwest (Table 1). Absolute incidence was greater among MSM and non-MSM men with prior opportunistic illnesses (particularly 2 illnesses), low nadir CD4 cell counts, and comorbid conditions (Supplementary Table 10, available online); for instance, incidence was 186 of 100 000 among MSM with diabetes in the United States (Table 2).

Table 1.

Absolute anal cancer incidence, relative risk compared with people without HIV, and calendar trends in anal cancer incidence from 2000 to 2016, NA-ACCORD

Characteristic	Incidence rate (95% CI)a	Relative risk vs people without HIV (95% CI)	Calendar trends
			Segment	Annual percentage change (95% CI)
North America
Overall	61.4 (56.4 to 66.6)	NRb	2000-2016	‒2.0% (‒3.9% to 0.1%)
Country
United States	60.4 (55.4 to 65.8)	27.5 (25.8 to 29.3)	2000-2016	‒2.2% (‒4.4% to ‒0.1%)
Canada	99.2 (69.7 to 139.0)	NRb	2000-2016	‒1.3% (‒6.5% to 4.2%)
US region
Northeast	29.3 (21.8 to 38.7)	15.9 (14.5 to 17.4)	2000-2016	0.4% (‒6.0% to 7.4%)
West	79.3 (68.7 to 90.9)	38.5 (36.5 to 40.7)	2000-2016	‒3.8% (‒6.5% to ‒0.9%)
Midwest	89.7 (71.6 to 110.9)	47.4 (45.0 to 49.9)	2000-2016	‒1.9% (‒7.2% to 3.7%)
South	56.6 (64.7 to 49.2)	26.9 (25.2 to 28.8)	2000-2016	‒2.7% (‒6.4% to 1.0%)

a

Incidence rates per 100 000 population (age standardized using the age distribution of people living with HIV in 2010). CI = confidence interval; NA-ACCORD = North American AIDS Cohort Collaboration on Research and Design; NR = not reported.

b

Not reported, given that the absolute age-stratified incidence rates among the general population in Canada were not available.

Table 2.

Absolute anal cancer incidence and relative risk (estimated using multivariable regression) among MSM and non-MSM men living with HIV, NA-ACCORD

Characteristic	MSM		Non-MSM men
	Incidence rate (95% CI)a	Incidence rate ratio (95% CI)b	Incidence rate (95% CI)a	Incidence rate ratio (95% CI)b
US region
Northeast	25.3 (10.4 to 49.6)	1 [Referent]	36.1 (26.6 to 49.1)	1 [Referent]
West	86.1 (69.1 to 106.2)	3.8 (1.9 to 7.6)	82.1 (68.1 to 98.7)	1.8 (1.2 to 2.6)
Midwest	137.7 (82.1 to 218.1)	4.1 (1.9 to 8.9)	90.5 (71.4 to 115.2)	2.4 (1.6 to 3.6)
South	105.5 (79.5 to 139.7)	4.7 (2.3 to 9.4)	50.7 (42.8 to 60.1)	1.3 (0.9 to 1.8)
Age, y
<35	43.1 (40.7 to 45.7)	1 [Referent]	33.9 (31.7 to 36.6)	1 [Referent]
35-44	86.9 (83.4 to 90.5)	1.6 (1.1 to 2.4)	67.0 (64.4 to 69.7)	1.5 (1.0 to 2.4)
45-54	104.9 (97.0 to 113.6)	1.9 (1.3 to 2.9)	64.1 (60.6 to 67.8)	1.5 (0.9 to 2.3)
≥55	88.4 (79.0 to 99.2)	1.6 (0.9 to 2.9)	62.9 (59.3 to 66.8)	1.5 (0.9 to 2.4)
Race and ethnicity
Hispanic	88.7 (52.5 to 146.8)	0.8 (0.5 to 1.3)	54.5 (39.6 to 73.5)	0.8 (0.5 to 1.2)
Non-Hispanic Black	94.4 (70.0 to 126.7)	1.1 (0.8 to 1.6)	42.3 (35.2 to 51.1)	0.5 (0.4 to 0.7)
Non-Hispanic White	83.0 (67.7 to 100.9)	1 [Referent]	85.5 (74.7 to 98.6)	1 [Referent]
Otherc	65.1 (24.6 to 136.4)	0.5 (0.2 to 1.1)	16.2 (6.5 to 31.6)	0.2 (0.1 to 0.6)
No. of opportunistic illnessesd
0	63.2 (50.7 to 78.2)	1 [Referent]	47.5 (41.6 to 54.4)	1 [Referent]
1	139.4 (101.7 to 190.3)	1.5 (1.1 to 2.1)	46.6 (32.5 to 66.6)	0.8 (0.5 to 1.1)
2	162.4 (116.9 to 220.9)	1.3 (0.9 to 1.9)	124.6 (104.1 to 151)	1.7 (1.3 to 2.2)
Nadir CD4 cell count, cells/mm3d
<100	166.2 (131.7 to 207.6)	4.0 (2.5 to 6.3)	96.7 (82.0 to 114.8)	2.5 (1.8 to 3.7)
100-200	101.3 (69.7 to 143.2)	2.5 (1.6 to 4.2)	67.1 (54.3 to 83.7)	2.0 (1.3 to 2.9)
200-349	49.9 (33.9 to 71.3)	1.5 (0.9 to 2.6)	50.1 (40.0 to 62.8)	1.4 (0.9 to 2.1)
≥350	40.8 (25.9 to 61.3)	1 [Referent]	34.1 (25.4 to 45.5)	1 [Referent]
Comorbidityd
No comorbidity	58.4 (41.3 to 83.0)	1 [Referent]	34.5 (25.0 to 47.0)	1 [Referent]
Chronic kidney disease	86.6 (72.0 to 105.5)	0.8 (0.4 to 1.5)	69.4 (60.5 to 79.4)	1.1 (0.6 to 1.8)
Diabetes	186.1 (141.5 to 246.1)	1.9 (1.1 to 3.2)	64.9 (38.1 to 84.0)	1.1 (0.7 to 1.5)
Cardiovascular disease	115.0 (93.9 to 140.7)	0.9 (0.4 to 2.0)	73.6 (64.5 to 84.7)	1.4 (0.7 to 2.7)
Hypertension	121.6 (97.8 to 151.8)	1.0 (0.5 to 2.1)	70.0 (59.1 to 81.7)	1.1 (0.6 to 2.1)
Comorbidity burdend
0	58.4 (41.3 to 83.0)	1 [Referent]	34.5 (25.0 to 47.0)	1 [Referent]
1	65.5 (47.6 to 89.1)	1.2 (0.6 to 2.5)	60.2 (48.1 to 75.7)	1.2 (0.6 to 2.2)
2	117.7 (84.7 to 163.1)	1.8 (0.5 to 6.4)	81.7 (65 to 102.3)	1.4 (0.5 to 4.0)
≥3	122.5 (96.3 to 156.9)	1.6 (0.2 to 11.7)	72.7 (49.6 to 87.8)	1.0 (0.2 to 5.0)

a

Incidence rates per 100 000 population (age standardized using the age distribution of people living with HIV in 2010). CI = confidence interval; MSM = men who have sex with men; NA-ACCORD = North American AIDS Cohort Collaboration on Research and Design.

b

Adjusted incidence rate ratios were derived from multivariable logistic regression models simultaneously adjusted for age group, race and ethnicity, number of opportunistic illnesses, nadir CD4 cell count, comorbidity, and comorbidity burden.

c

Other included Asian, American Indian or Alaska Native, multiracial, and other unspecified.

d

Number of opportunistic illnesses, nadir CD4 cell count, comorbidity, and comorbidity burden were time-updated characteristics.

When further stratified by age, for people with prior opportunistic illnesses, particularly with 2 illnesses, incidence was 100 or more per 100 000 (≥2-fold risk compared with people who had no opportunistic illnesses) and stable across age groups (Figure 1; Supplementary Table 11, available online). In contrast, incidence was lower (<25/100 000) among people with HIV younger than 35 years of age with no opportunistic illnesses. Similarly, across age groups, incidence was elevated among people with HIV who had a nadir CD4 cell count below 200 cells/mm³.

Figure 1.

Anal cancer incidence according to age and (A) prior opportunistic illness diagnosis, (B) number of opportunistic illnesses, and (C) nadir CD4 cell count. Absolute anal cancer incidence and accompanying error bars show 95% confidence intervals.

Across age groups, cumulative risk was greater for MSM than for non-MSM men (Figure 2; A). When stratified according to the number of opportunistic illnesses, however, for the group younger than 35 years of age, the risk at 10 and 15 years among non-MSM men with 2 opportunistic illnesses exceeded that among MSM with no opportunistic illnesses (Figure 2, B; Supplementary Table 12, available online). For the groups aged 35 to 54 years and 55 years and older, cumulative risks among non-MSM men with 2 opportunistic illnesses were similar to MSM with no opportunistic illnesses. Additional findings for stratification by nadir CD4 cell count are presented in Figure 2, C.

Figure 2.

Cumulative incidence of anal cancer among MSM with HIV and non-MSM men with HIV (A) overall, (B) according to the number of opportunistic illnesses, and (c) nadir CD4 cell count, measured since time at cohort entry for people with HIV younger than 35 years of age, 35 to 54 years of age, and 55 years of age and older. MSM = men who have sex with men.

^a Cumulative incidence estimates for people with HIV with no opportunistic illness were measured from the time since cohort entry. For people with HIV with 1 or 2 opportunistic illnesses, the cumulative incidence was measured from the time since diagnosis of the opportunistic illness.

^b Cumulative incidence estimates were measured from the time since the respective CD4 cell counts were first observed.

In the United States, during 2000-2016, anal cancer incidence decreased 2.2% per year (95% CI = ‒4.4% to ‒0.1%) (Table 1; Supplementary Table 13, available online). In contrast, in Canada, no statistically significant change in incidence was observed (‒1.3% per year, 95% CI = ‒6.5% to 4.2%). In the United States, incidence declined significantly in the Western region (‒3.8% per year, 95% CI = ‒6.5% to ‒0.9%). The annual percentage change was negative but not significant for other regions.

In the United States, the relative risk compared with people without HIV was 27.5 (95% CI = 25.8 to 29.3) (Supplementary Table 14, available online) and 44.7 for MSM (95% CI = 42.5 to 47.1). The relative risk varied greatly by US region: 15.9 (95% CI = 14.5 to 17.4) in the Northeast, 26.9 (95% CI = 25.2 to 28.8) in the South, 38.5 (95% CI = 36.5 to 40.7) in the West, and 47.4 (95% CI = 45.0 to 49.9) in the Midwest. Among MSM with HIV in the Midwest, relative risk reached 96.2 (95% CI = 92.4 to 100.2); risk was lowest in the Northeast (20.3, 95% CI = 18.5 to 22.3). The relative risk decreased with increasing age and varied by race and ethnicity (Supplementary Table 14, available online).

Specific to the United States, relative risk (95% CI) from multivariable Poisson regression analysis for MSM and non-MSM men is presented in Table 2. For both groups, the risk was lower in the Northeast than in the other regions. Risk increased after 35 years of age. Risk was elevated among White non-MSM men with HIV compared with other racial and ethnic groups, but differences according to race and ethnicity were not significant among MSM with HIV. Anal cancer risk increased with an increasing number of opportunistic illnesses and decreasing levels of nadir CD4 cell count. The risk was 2-fold higher among MSM with diabetes than among those individuals with no comorbidity.

To project future anal cancer incidence, we first used our microsimulation model to project demographic trends for American people with HIV. The model output showed that the number of adults with HIV in the United States is increased from 1.01 million in 2011 to 1.12 million in 2016 and is expected to begin to plateau in 2028 (at 1.2 million) (Figure 3, A). The proportion of people with HIV 55 years of age and older will contribute to 47.6% of the HIV burden in 2035 (from 28.8% in 2016). The distribution by transmission risk group (MSM, other men, and women) and people with a history of AIDS is expected to remain the same (Figure 3, B and C), with the majority (60%-65%) of the cases being MSM with HIV and people with HIV who had no prior AIDS diagnosis.

Figure 3.

Projected estimates and proportional distribution of people with HIV in the United States from 2011 to 2035 by (A) age group, (B) risk group (MSM, non-MSM men, and women), and (C) prior AIDS diagnosis (opportunistic illnesses or nadir CD4 cell count <200 cells/mm³) status. MSM = men who have sex with men.

The annual number of anal cancer cases in people with HIV is expected to decrease from 770 (uncertainty interval = 625-918) in 2016 to nearly 574 (uncertainty interval = 440-726) in 2035 (Figure 4, A). Particularly noteworthy is the anticipated shift in the age distribution: The proportion of patients 55 years of age and older, which was 39% in 2016, is projected to rise to 56% by 2035 (Figure 4, A). Nearly 80% of the anal cancer burden among those with HIV will continue to occur among MSM with HIV (Figure 4, B). People with prior AIDS diagnosis will continue to incur 60% or more of the anal cancer burden among people with HIV (Figure 4, C), of which more than 70% will occur among MSM with AIDS. Additional findings, including incidence rate projections and a scenario that assumes no change in incidence rates, are presented in Supplementary Figures 10-14 (available online).

Figure 4.

Projected estimates and proportional distribution of anal cancer cases among people with HIV in the United States from 2011 to 2035 by (A) age group, (B) risk group (MSM, non-MSM men, and women), and (C) prior AIDS diagnosis status (opportunistic illnesses or nadir CD4 cell count <200 cells/mm³) status. MSM = men who have sex with men.

Discussion

During 2001-2016, anal cancer incidence decreased 2.2% per year among people with HIV in the United States, with a notably significant 3.7%-per-year decline in the Western region. In contrast, incidence was stable and relatively greater (nearly 2-fold higher) in Canada. In the United States, a large geographic variation in incidence was observed, with the highest absolute and relative risk in the Midwest region, while the risk was relatively lower among people with HIV in the Northeast region. Over the next decade, the anal cancer burden among people with HIV is expected to decline. Of note, the age distribution of anal cancer will shift toward older (≥55 years of age) people with HIV, with a high proportion (>70%) of cases expected to occur among MSM with HIV and people with HIV with prior AIDS diagnoses.

The declining anal cancer trend among people with HIV in the United States likely reflects immune preservation from the wide use of ART in the past decade. Our observed patterns may also reflect the early adoption and implementation of screening in some regions. Regional guidelines exist from the New York Department of Health AIDS Institute, health-care professionals in California have been screening for anal cancer since the 1990s (4,24). Consistent with our findings, a prior ecological study from Kaiser Permanente Northern California demonstrated a reduction in anal cancer incidence after the introduction of screening (25). Higher absolute incidence and relatively slower decline in incidence in the Midwest and Southeast regions likely reflect underlying variations in the HIV epidemic and care as well as suboptimal screening infrastructure and resultant differences in screening practices. Particularly, the South is known to have a greater proportion of undiagnosed HIV and late HIV diagnoses, whereas screening infrastructure is sparse in midwestern and southern states (4). Indeed, a recent study showed that in 2019, only 1.4% of people with HIV in the midwestern and 3% in southern states were screened in the past 12 months using anal cytology (as opposed to 9% in western and 5.2% in northeastern states) (20). Our observed increased risk in Canada may reflect variations in ART use or differences in screening practices. Early adoption and wide use of ART also occurred in Canada over the past decade, and ART adherence is similar to that observed in the United States (26); thus, a difference in immune preservation is unlikely. Anal cancer screening practices may have differed; however, the absence of data on screening utilization in Canada limits our ability to make any inference regarding incidence patterns in association with screening. Future research is needed to elucidate factors underlying anal cancer incidence patterns in Canada. Of note, despite our attempt to explain regional patterns, we caution against the overinterpretation of these findings.

The geographic variations that we observed also have implications to guide screening practices. To translate screening benefits into cancer prevention at the population level, it will be imperative to have a high uptake of screening and diagnostic follow-up (high-resolution anoscopy) as well as timely detection and treatment of precancerous lesions. The regional variations in risk may imply that if equitable screening capacity is not achieved in regions with higher risk, our observed geographical disparities may magnify in the future.

Our findings have implications for guiding risk stratification for informing anal cancer screening recommendations. Anal cancer screening recommendations for people with HIV in the United States have rapidly evolved, from expert opinion–based guidance to evidence-based consensus guidelines (18,19). Emerging anal cancer screening recommendations have defined incidence thresholds of 25 or more or 50 or more per 100 000 population or relative risk 10-fold or higher than the general population as benchmarks to identify risk and age groups for anal cancer screening (18,19). Based on these thresholds and prior knowledge of absolute risk, screening is not recommended among people with HIV younger than 35 years of age (17,19). Our findings suggest that in people with HIV younger than 35 years of age with prior opportunistic illnesses or nadir CD4 cell counts below 200 cells/mm³, absolute anal cancer incidence exceeds 50 per 100 000 population, reaching more than 125 per 100 000 population among those individuals with 2 illnesses, implying that young people with HIV who have these risk factors meet the benchmark for screening consideration. Furthermore, the cumulative risk among MSM younger than 35 years of age with 2 opportunistic illnesses was nearly identical to that among MSM 35 to 54 years of age with 2 opportunistic illnesses. Our cumulative risk analysis also implies that opportunistic illnesses and nadir CD4 cell count are equally important or more significant predictors of anal cancer risk as the HIV risk group (MSM, non-MSM men), implying opportunities to consider these predictors to guide risk stratification for screening use.

This study is the first to report an association between anal cancer risk and opportunistic illnesses and comorbidities. The role of immunosuppression in accelerating anal carcinogenic progression is known (10); however, elevated risk in association with a comorbid condition such as diabetes or increasing comorbidity burden introduces other potential mechanisms of action (eg, inflammation) that remain unstudied among people with HIV. More research is needed to elucidate anal cancer risk in association with comorbidities. These findings also have important implications for screening because the harms and benefits of screening and downstream screening outcomes may differ in the context of comorbid conditions (eg, attributable to mortality from other processes).

If recent patterns in incidence trends persist, the anal cancer burden among people with HIV is projected to decline. An increasing proportion of cases will occur, however, among people with HIV 55 years of age and older. The shift in anal cancer burden toward older individuals is reflective of greater absolute incidence among older people with HIV, increasing prevalence of older people with HIV with prior AIDS diagnoses, and their improved life expectancy. In contrast, the declining burden among young people with HIV is suggestive of decreased HIV incidence among young people, particularly individuals with prior AIDS diagnoses. Furthermore, although people with prior AIDS diagnoses represent 40% of individuals with HIV, more than 75% of anal cancers diagnosed among people with HIV will continue to occur among this group. We also project that nearly 80% of anal cancer cases with HIV will occur among MSM. The high burden of anal cancer among these groups highlights the importance of HIV prevention and early HIV care as well as adherence to ART and screening to potentially mitigate the HIV-attributable anal cancer burden.

Our study has certain limitations. First, NA-ACCORD is representative of patients with HIV linked to health care in North America; therefore, inferences made in this study about people with HIV are limited to individuals engaged in HIV care. Second, the geographic patterns are limited to cohorts represented in NA-ACCORD and do not reflect complete population coverage of people with HIV in specific regions. Similarly, findings specific to Canada may not reflect complete population coverage because these data were limited to participating cohorts from eastern and western Canada. Additional research is needed to understand the generalizability of our findings. Third, the projection model relies on overall calendar trends and does not capture trends specific to risk groups (eg, MSM, non-MSM men, and women); however, because trends among major risk groups (ie, MSM and non-MSM men) were similar in magnitude and because women represent a small fraction of HIV and anal cancer cases, the impact of this assumption on calendar trends is likely minimal. Fourth, the risk estimation and incidence projections were limited to the United States because of the robust data availability on cancer incidence among the general population available through the US Cancer Registry Databases, a larger sample size that enabled us to stratify incidence data among people with HIV, and robust availability of the HIV prevalence data from the HIV Surveillance System. The absence of similar data for Canada precluded the ability to derive relative risk and project future burden. Fifth, the mathematical model does not directly capture the impact of human papillomavirus vaccination and screening on anal cancer burden. Because anal cancer is a disease that occurs in older-aged individuals, however, the bulk of the impact of human papillomavirus vaccination on reducing anal cancer incidence will unlikely be seen until 2035 and thus unlikely to affect our projections. In contrast, an early impact of screening on anal cancer incidence reduction is possible, dependent on the magnitude of screening uptake and subsequent treatment of anal precancerous lesions in the future.

In summary, anal cancer incidence declined among people with HIV in the United States, particularly in the western region. Absolute anal cancer incidence was nearly 2-fold greater in Canada than in the United States. Geographic variation in anal cancer risk in the United States is likely the result of underlying differences in the HIV epidemic and screening practices. If the current trends in incidence persist, the anal cancer burden among people with HIV is expected to decline in the next decade. Most individuals diagnosed with anal cancer, however, will occur among people with HIV 55 years of age and older. In addition, MSM with HIV and people with HIV with prior AIDS diagnoses will bear most of the anal cancer burden in the United States among people with HIV. Our findings highlight the need for improvement in HIV and anal cancer prevention to reduce disparities in anal cancer risk among people with HIV.

Data availability

NA-ACCORD welcomes interested investigators to collaborate with us for use of our data. Please visit https://naaccord.org/ for additional information.

Author contributions

Ashish Deshmukh, PhD, MPH (Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Software; Supervision; Validation; Visualization; Writing—original draft; Writing—review & editing); Yueh Yun Lin, PhD (Data curation; Formal analysis; Investigation; Visualization; Writing—review & editing); Haluk Damgacioglu, PhD (Data curation; Formal analysis; Investigation; Methodology; Validation; Visualization; Writing—review & editing); Meredith Shiels, PhD (Conceptualization; Investigation; Methodology; Validation; Visualization; Writing—review & editing); Sally Coburn, PhD (Investigation; Methodology; Writing—review & editing); Raynell Lang, MD (Investigation; Writing—review & editing); Keri N. Althoff, PhD (Investigation; Methodology; Writing—review & editing); Richard Moore, MD (Investigation; Writing—review & editing); Michael Silverberg, PhD (Investigation; Writing—review & editing); Alan Nyitray, PhD (Investigation; Writing—review & editing); Jagpreet Chhatwal, PhD (Investigation; Writing—review & editing); Kalyani Sonawane, PhD (Conceptualization; Formal analysis; Methodology; Writing—review & editing); Keith Sigel, MD, PhD, MPH (Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Supervision; Validation; Visualization; Writing—original draft; Writing—review & editing).

Funding

This work was supported by the following National Institutes of Health grants: R01CA232888, U01AI069918, F31AI124794, F31DA037788, G12MD007583, K01AI093197, K01AI131895, K23EY013707, K24AI065298, K24AI118591, K24DA000432, KL2TR000421, N01CP01004, N02CP055504, N02CP91027, P30AI027757, P30AI027763, P30AI027767, P30AI036219, P30AI050409, P30AI050410, P30AI094189, P30AI110527, P30MH62246, R01AA016893, R01DA011602, R01DA012568, R01AG053100, R24AI067039, R34DA045592, U01AA013566, U01AA020790, U01AI038855, U01AI038858, U01AI068634, U01AI068636, U01AI069432, U01AI069434, U01DA036297, U01DA036935, U10EY008057, U10EY008052, U10EY008067, U01HL146192, U01HL146193, U01HL146194, U01HL146201, U01HL146202, U01HL146203, U01HL146204, U01HL146205, U01HL146208, U01HL146240, U01HL146241, U01HL146242, U01HL146245, U01HL146333, U24AA020794, U54GM133807, UL1RR024131, UL1TR000004, UL1TR000083, UL1TR002378, Z01CP010214 and Z01CP010176. It received further funding from the Intramural Research Program of the National Cancer Institute; contracts CDC-200-2006-18797 and CDC-200-2015-63931 from the US Centers for Disease Control and Prevention; contract 90047713 from the US Agency for Healthcare Research and Quality; contract 90051652 from the US Health Resources and Services Administration; the Grady Health System; grants CBR-86906, CBR-94036, HCP-97105, and TGF-96118 from the Canadian Institutes of Health Research; and the Ontario Ministry of Health and Long Term Care as well as the government of Alberta, Canada. Additional support was provided by the National Institute of Allergy And Infectious Diseases; National Cancer Institute; National Heart, Lung, and Blood Institute; Eunice Kennedy Shriver National Institute of Child Health & Human Development; National Human Genome Research Institute; National Institute for Mental Health; National Institute on Drug Abuse; National Institute on Aging; National Institute of Dental & Craniofacial Research; National Institute of Neurological Disorders and Stroke; National Institute of Nursing Research; National Institute on Alcohol Abuse and Alcoholism; National Institute on Deafness and Other Communication Disorders; and National Institute of Diabetes and Digestive and Kidney Diseases. These data were collected by cancer registries participating in the National Program of Cancer Registries of the Centers for Disease Control and Prevention. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Centers for Disease Control and Prevention.

Conflicts of interest

Drs Deshmukh and Sonawane have consulted for Value Analytics Labs on unrelated projects. Dr Deshmukh was a member of the scientific advisory board on unrelated projects for Merck.

Acknowledgements

Role of the Funders/Support: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.