Association of HIV Status With Outcomes of Anal Squamous Cell Carcinoma in the Era of Highly Active Antiretroviral Therapy

Abstract

This study uses data from the Veterans Affairs database to examine the association of HIV status with outcomes in patients with anal cancer.

The treatment of anal cancer in patients who are seropositive for HIV (HIV-positive) has been controversial for more than 3 decades. Current treatment guidelines recommend that healthy HIV-positive patients receive standard fluorouracil and mitomycin C concurrent chemoradiotherapy, but some investigators question whether these patients experience more severe toxic effects or worse long-term outcomes that might necessitate a modified treatment approach. This study investigates the influence of HIV status on patient outcomes in a national cohort of US veterans.

Methods

This study was approved by the institutional review board of the Veterans Administration (VA) San Diego, and the need for informed consent was waived. The data were not deidentified because they were accessed from within the secure VA Informatics and Computing Infrastructure, which includes patient-level data collected for the use of VA researchers. From a nationwide VA database, we identified patients with clinical stage I through III anal squamous cell carcinoma diagnosed between 2000 and 2015 who were treated with definitive chemoradiotherapy. Outcomes included completion of chemotherapy, radiotherapy treatment breaks, acute toxic effects, cause-specific (tumor recurrence/progression or treatment toxic effects) ostomy rates, and cause-specific mortality. We used multivariable logistic regression models to analyze completion of chemotherapy, radiotherapy treatment breaks, and acute toxic effects, each on a complete-case basis. All-cause mortality was analyzed using the Cox proportional hazards regression model; ostomy rates and cause-specific mortality were analyzed with competing risk regression models. Covariates in each model included HIV status, pretreatment positron-emission tomography, tumor stage, node stage, comorbidity, age, race, year of diagnosis, intensity-modulated radiotherapy, mitomycin C chemotherapy, and history of cancer. A 2-sided P < .05 defined statistical significance.

Results

The sample of 833 included 150 (18.0%) HIV-positive and 683 (82.0%) HIV-negative patients. Patients with HIV were more likely to be male (149 of 150 patients [99.3%] vs 617 of 683 [90.3%]; P < .001), younger (mean [SD] age, 55 [8.6] vs 63 [9.6] years; P < .001), and black (47 of 150 patients [31.3%] vs 63 of 683 [9.2%]; P < .001). There were no differences in tumor/node stage, comorbidity, chemotherapy regimen, or radiation dose. Among HIV-positive patients, the median pretreatment CD4 lymphocyte count was 370/µL (interquartile range, 205-543/µL) (to convert to cells ×10⁹/L, multiply by 0.001), and 91% of patients had received highly active antiretroviral therapy (HAART) in the 6 months before cancer treatment.

Patients with HIV were as likely as HIV-negative patients to miss a second cycle of any chemotherapy (odds ratio [OR], 1.38; 95% CI, 0.77-2.47; P = .27), but were more likely to miss their second cycle of mitomycin C (OR, 2.03; 95% CI, 1.20-3.44; P = .008; Table 1). Fifty-eight HIV-positive patients (46%) experienced radiotherapy treatment breaks of 5 days or more, compared with 176 (33%) of HIV-negative patients (OR, 1.66; 95% CI, 1.07-2.56; P = .02), but there was no difference in breaks of 10 days or more. Patients with HIV were more likely to experience grade 3 or 4 hematologic toxic effects (OR, 2.18; 95% CI, 1.45-3.28; P < .001). Patients with HIV were more frequently hospitalized within 90 days of treatment for acute toxic effects (OR, 1.75; 95% CI, 1.16-2.66; P = .008), in particular for hematologic (OR, 2.20; 95% CI, 1.33-3.64; P = .002) but not gastrointestinal (OR, 1.35; 95% CI, 0.75-2.44; P = .32; Table 1) toxic effects. The long-term ostomy rate did not differ between HIV-positive and HIV-negative patients. We found increased noncancer mortality for the HIV-positive cohort, although there were no differences between cohorts in cancer-specific or all-cause mortality (Table 2).

Table 1. Association of HIV Status With Short-term Outcomes in Patients With Anal Squamous Cell Carcinoma^a.

Outcome	Frequency of Events, No. (%)		Multivariable Analysis
	HIV Positive (n = 150)	HIV Negative (n = 683)	OR (95% CI)	P Valueb
Treatment completion
No second chemotherapy cycle	21 (16.5)	84 (14.3)	1.38 (0.77-2.47)	.27
No second dose of mitomycin C	45 (45.5)	162 (34.6)	2.03 (1.20-3.44)	.008
Radiotherapy treatment interruption, d
≥5	58 (46.0)	176 (32.8)	1.66 (1.07-2.56)	.02
≥10	30 (23.8)	84 (15.6)	1.48 (0.88-2.51)	.14
Acute toxic effects
Grade 3 or 4 hematologic toxic effectsc	84 (58.7)	245 (39.7)	2.18 (1.45-3.28)	<.001
90-d Hospitalization
Any treatment toxic effects	54 (36.2)	174 (26.3)	1.75 (1.16-2.66)	.008
Hematologic toxic effects	34 (22.8)	78 (11.8)	2.20 (1.33-3.64)	.002
Gastrointestinal toxic effects	19 (12.8)	76 (11.5)	1.35 (0.75-2.44)	.32

Abbreviation: OR, odds ratio.

^aThis table demonstrates the influence of HIV status on treatment completion and acute toxic effects. The ORs represent the results of a multivariable logistic regression analysis, with values greater than 1 indicating an increased risk among patients with HIV. Short-term was defined as 90-day toxic effects and treatment completion outcomes.

^bAll P values are 2-sided.

^cGrades 3 and 4 hematologic toxic effects were defined as hemoglobin level, <8.0 g/dL (to convert to grams per liter, multiply by 10) or requiring transfusion; platelet count, <50 ×10³/µL (to convert to ×10⁹/µL, multiply by 1); leukopenia or white blood cell count, <2000/µL (to convert to ×10⁹/L, multiply by 0.001); or neutrophils, <1000/µL (to convert to ×10⁹/L, multiply by 0.001).

Table 2. Association of HIV Status With Long-term Outcomes in Patients With Anal Squamous Cell Carcinoma^a.

Outcome	5-y Cumulative Incidence, %		Multivariable Analysis
	HIV Positive (n = 150)	HIV Negative (n = 683)	HR/SDHR (95% CI)b	P Valuec
Ostomy
All-cause	14	13	1.21 (0.74-1.97)	.45
Treatment-related	3.3	3.3	0.95 (0.35-2.59)	.92
Tumor-related	11	10	1.34 (0.76-2.38)	.32
Mortality
Cancer-specific	13	19	0.87 (0.52-1.47)	.61
All-cause	41	38	1.25 (0.93-1.67)	.14
Noncancer	26	17	1.54 (1.03-2.30)	.04

Abbreviations: HR, hazard ratio; SDHR, subdistribution hazard ratio.

^aLong-term was defined as mortality and ostomy placement for tumor recurrence or late toxic effects from radiotherapy.

^bSDHR reflects the results of multivariable Fine-Gray competing risk regression model for the ostomy and cause-specific mortality outcomes, and HR reflects the results of the Cox proportional hazards regression for all-cause mortality. Values greater than 1 indicate an increased risk among HIV-positive patients.

^cP values are 2-sided.

Discussion

Studies evaluating the association of HIV status with cancer outcomes have produced mixed results, with some showing increased toxic effects or inferior long-term outcomes and others showing no difference. Earlier studies have been limited by smaller sample sizes, single-institution settings, or cohorts with uncontrolled HIV drawn before the availability of HAART. It is not clear that the higher rates of acute toxic effects observed in those studies still apply to a post-HAART cohort with better-controlled HIV disease.

We demonstrate that patients with HIV have a higher risk than HIV-negative patients for acute treatment-related toxic effects, especially hematologic toxic effects. However, this finding does not translate to inferior long-term outcomes, and HIV-positive patients showed survival and ostomy placement rates equivalent to those of their HIV-negative counterparts. These data point to the need to optimize treatment in this population to decrease or better manage acute toxic effects. Our results also paint an optimistic picture of the long-term prognosis for HIV-positive patients, reflecting the improvements in HIV disease control and supportive care for this vulnerable population.