Pooled analysis of AIDS malignancy consortium trials evaluating rituximab plus CHOP or infusional EPOCH chemotherapy in HIV-associated non-Hodgkin lymphoma




Improved outcomes have recently been reported for rituximab (R) plus rituximab plus infusional etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin (R-EPOCH) chemotherapy in patients with human immunodeficiency virus (HIV)-associated, aggressive B-cell, non-Hodgkin lymphoma (NHL). The objective of the current analysis was to assess whether patient selection or other factors contributed to this improvement and to identify patients who are at the greatest risk for lethal toxicity.


The authors performed a pooled analysis of 2 consecutive trials that included 150 patients with HIV-associated NHL who received either R-CHOP (n = 99; Acquired Immunodeficiency Syndrome [AIDS] Malignancy Consortium Trial 010 [AMC010]) or R-EPOCH (n = 51; AMC034). Age-adjusted International Prognostic Index (aaIPI), CD4 count at lymphoma diagnosis (<100/μL vs ≥100/μL), and treatment (R-CHOP vs R-EPOCH) were included as variables in a multivariate logistic regression model for complete response (CR) and in a Cox proportional hazards regression models for event-free survival (EFS) and overall survival (OS).


Features that were associated significantly with an improved CR rate and improved EFS and OS included a low aaIPI score and a baseline CD4 count ≥100/μL. When the analysis was adjusted for aaIPI and CD4 count, patients who received concurrent R-EPOCH had improved EFS (hazard ratio [HR] 0.40; 95% confidence intervals [CI], 0.23, 0.69; P < .001) and OS (HR, 0.38; 95% CI, 0.21, 0.69; P < .01). Treatment-associated death occurred significantly more often in patients with CD4 counts <50/μL (37% vs 6%; P < .01).


The current analysis provided additional level 2 evidence supporting the use of concurrent R-EPOCH in patients with HIV-associated lymphoma and a CD4 count >50/μL, and the results support the design of an ongoing phase 3 trial comparing concurrent R-EPOCH with R-CHOP in immunocompetent patients with diffuse large B-cell lymphoma (National Clinical Trial no. NCT00118209). Cancer 2012. © 2011 American Cancer Society.


The incidence of aggressive non-Hodgkin lymphoma (NHL) is up to approximately 600-fold greater in individuals with human immunodeficiency virus (HIV) infection. Despite a decrease in the incidence of HIV-associated NHL since the advent of combination antiretroviral therapy (cART), NHL remains the most common HIV-associated malignancy in the United States.1–5

Rituximab (Rituxan; Genentech, Inc., San Francisco, Calif) is a monoclonal antibody directed against the cluster of differentiation 20 (CD20) antigen present on malignant and normal B lymphocytes. Rituximab is a highly effective treatment for immunocompetent patients with B-cell lymphoma, whether used alone in selected patients with low-grade B-cell lymphoma or in combination with chemotherapy for patients with low-grade and aggressive B-cell lymphoma.6, 7 It is approved by the US Food and Drug Administration for the treatment of CD20-positive NHL. It has been demonstrated that the addition of rituximab to standard combined cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) chemotherapy or to CHOP-like regimens significantly improves complete response (CR) and overall survival (OS) rates in patients with diffuse large B-cell lymphoma (DLBCL) in several trials.8-10 For HIV-associated lymphoma only 1 phase 3 study with rituximab has been reported. The Acquired Immunodeficiency Syndrome (AIDS) Malignancy Consortium (AMC) compared rituximab plus CHOP (R-CHOP) with CHOP alone and observed that the addition of rituximab resulted in fewer deaths from lymphoma (14% vs 29%; P < .05) offset by a higher infectious death rate (14% vs 2%; P < .05) in the R-CHOP arm, resulting in comparable rates of CR rates and OS (AMC Trial 010 [AMC010]).11 Other phase 2 trials in HIV-associated lymphoma have documented safety and efficacy for R-CHOP,12, 13 suggesting that the addition of rituximab to chemotherapy should be considered the standard of care for most HIV-infected patients with B-cell lymphomas.14

Several phase 2 trials have suggested that infusional cytotoxic therapy may be more effective than CHOP in patients with aggressive B-cell NHL, including immunocompetent15, 16 and HIV-positive17,18 patients, whether or not they receive rituximab concurrently with chemotherapy. In AMC034, patients with HIV-associated, aggressive B-cell NHL were randomized to receive infusional etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin (EPOCH) chemotherapy either concurrently with rituximab (R-EPOCH) or followed sequentially by rituximab given weekly for 6 weeks after the completion of chemotherapy.19 The CR rate was 69% in the concurrent arm (95% confidence interval [CI], 56%-79%) and 53% in the sequential arm (95% CI, 41%-64%). In that randomized “pick-the-winner” phase 2 trial design, the null hypothesis that the CR rate was 50% versus the alternative of 75% was rejected for the concurrent arm in favor of the alternative (P = .005) and was accepted for the sequential arm (P = .39), suggesting that concurrent R-EPOCH is more effective than sequential therapy. The objectives of the current pooled analysis, which included patients who received concurrent rituximab plus chemotherapy in AMC010 (R-CHOP) and AMC034 (R-EPOCH), were to determine whether the apparent benefit of infusional therapy persisted after adjustment of known prognostic covariates and to identify patients at high risk of lethal toxicity when receiving rituximab plus chemotherapy.


Description of Clinical Trials

We pooled data from the 2 consecutively performed AMC trials, including AMC01011 and AMC034.19 The characteristics of the trials are listed in Table 1 and include the periods during which the studies were performed, lymphoma therapy, antiretroviral therapy, and supportive care. Both trials included the same AMC sites and the same group of investigators. Both trials had similar eligibility criteria. HIV positive patients aged ≥18 years with an Eastern Cooperative Oncology Group performance status from 0 to 2 and adequate organ function were eligible if they had previously untreated, aggressive, CD20-positive B-cell NHL that was stage II or greater (or stage I with an elevated lactate dehydrogenase level). Both protocols included Pneumocystis jirovecii prophylaxis and the use of myeloid growth factors. Meningeal prophylaxis was administered at the discretion of the treating physician and was recommended for patients at high risk of central nervous system relapse, as defined previously.11, 19 No antibiotic prophylaxis for neutropenia was recommended in AMC010, whereas fluoroquinolones and fluconazole were added in AMC034. Antiretroviral therapy was mandatory for patients in AMC010 but was left at the treating investigator's discretion in AMC034.

Table 1. Treatment Regimens for AIDS Malignancy Consortium (AMC) Trials AMC010 and AMC034
  • Abbreviations: CHOP, cyclophosphamide, vincristine, doxorubicin, and prednisone; EPOCH, etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin; IV, intravenous; PO, orally; R-CHOP, rituximab plus CHOP; R-EPOCH, rituximab plus EPOCH.

  • a

    Rituximab was given 2 days (day −2) before starting the chemotherapeutic agents. Day 1 denotes the first day of chemotherapy.

Period of accrual1998-20022002-2006
Antilymphoma therapyR-CHOPR-EPOCH
 Rituximab375 mg/m2 IV before each CHOP cyclea375 mg/m2 IV before each EPOCH cyclea
 PrednisoneDaily 100 mg PO on D1-D5aDaily 60 mg/m2 PO on D1-D5a
 Cyclophosphamide (IV bolus over 60 min)750 mg/m2 IV on D1aStart at 187.5 mg/m2 (for CD4 counts <100/μL) or 375 mg/m2 (for CD4 counts ≥100/μL); escalate dose to 750 mg/m2 IV on D5 after completion of etoposide, vincristine, and doxorubicin infusiona
 EtoposideDaily 50 mg/m2 for 4 d as a continuous, 96-h, IV infusion on D1-D4
 Vincristine1.4 mg/m2 (maximum, 2.0 mg/m2) IV push on D1Daily 0.4 mg/m2 for 4 d as a continuous, 96-h, IV infusion on D1-D4
 Doxorubicin50 mg/m2 IV push on D1Daily 10 mg/m2 for 4 d as a continuous, 96-h, IV infusion on D1-D4
Meningeal prophylaxisAt discretion of local investigator; recommended if small, noncleaved histology; bone marrow, paranasal sinus, testicular, or epidural involvementAt discretion of local investigator; recommended if small, noncleaved histology; bone marrow, paranasal sinus, testicular, or epidural involvement
Concurrent antiretroviral therapyMandatoryAt discretion of treating physician
Infection prophylaxisPneumocystis jirovecii; antibacterial and antifungal not recommendedPneumocystis jirovecii; antifungal (fluconazole) and antibacterial (fluoroquinolone)
Granulocyte colony stimulating factorUpon completion of chemotherapy until neutrophil recoveryUpon completion of chemotherapy until neutrophil recovery

Statistical Analysis

We pooled data from 99 patients who received R-CHOP in AMC010 and 51 patients who received concurrent R-EPOCH in AMC034. Normally distributed data are presented as the mean ± standard deviation. For univariate associations of variables, means were compared using the Student t test for normally distributed data and, alternatively, the nonparametric Mann-Whitney U test. Proportions were assessed for associations using the Pearson chi-square test or the nonparametric Fisher exact test. Actuarial graphs for the outcomes event-free survival (EFS) (defined as time between registration and either relapse or progression of lymphoma or death from any cause) and OS were plotted according to the Kaplan-Meier method. The log-rank test was used to compare the treatment arms with respect to EFS and OS. A multivariate logistic regression model was used to estimate the odds ratio (OR) for the outcome complete response (CR) (defined as confirmed or unconfirmed complete response [CR/CRu] according to international response criteria20). Other variables that we included were the age-adjusted International Prognostic Index (aaIPI) (low-risk score [0-1] or high-risk score [2-3]), CD4 count (<100/μL or ≥100/μL), and treatment (R-CHOP or R-EPOCH). A Cox proportional hazards regression model was used to estimate the hazard ratio (HR) for the outcomes EFS and OS. Test assumptions for all statistical tests were evaluated, and we established that they were not violated. A P value < .05 was considered statistically significant and resulted in rejection of the null hypothesis; all statistical tests were 2-sided. The 95% CI was calculated whenever applicable. Statistical analyses were performed using the SAS statistical software package (version 9.2; SAS Institute Inc., Cary, NC).



The characteristics of patients from both trials who were included in this analysis are provided in Table 2. There were no significant differences in age, sex, CD4 count at lymphoma diagnosis, or histology between the study groups, although patients treated on AMC034 tended to have a higher median CD4 count. The most common histologic lymphoma subtype in both trials was DLBCL.

Table 2. Characteristics of the Patients on AIDS-Malignancy Consortium (AMC) Trials AMC010 and AMC034
 No. of Patients (%) 
CharacteristicAMC010: R-CHOP, N = 99AMC034: R-EPOCH, N = 51P
  • Abbreviations: AIDS, Acquired Immunodeficiency Syndrome; DLBCL, diffuse large B-cell lymphoma; IPI, International Prognostic Index; R-CHOP, rituximab plus cyclophosphamide, vincristine, doxorubicin, and prednisone; R-EPOCH, rituximab plus etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin; SD, standard deviation.

  • a

    P values are for differences in the median CD4 count between the 2 groups.

  • b

    For 11 patients in AMC010, the baseline CD4 count was missing.

  • c

    Other lymphoma subtypes included the following: in AMC010, Burkitt lymphoma (n = 8; 8%); high-grade lymphoma, not otherwise specified (n = 4; 4%); polymorphic B-cell lymphoma (n = 2; 2%); and, in 1 patient each (1%), mixed histology, primary effusion, and missing; in AMC034, Burkitt/Burkitt-like lymphoma (n = 16; 31%).

Age: Mean±SD, y43.5±8.342.6±8.4.52
Men90 (91)43 (84).28
Age-adjusted IPI   
 Low risk41 (41)16 (31) 
 High risk58 (59)35 (69).29
Baseline CD4 count, /μL   
 Median [range]130 [1-2457]181 [2-882].28a
 <5022 (22)8 (16) 
 <10036 (36)b16 (31) 
 ≥10052 (52)b35 (69) 
 DLBCL80 (81)35 (69).11
 Otherc19 (19)16 (31) 

Relation Between Treatment and Event-Free Survival and Overall Survival

We evaluated the relation between treatment and clinical outcomes without adjustment for CD4 count or aaIPI. For all patients, R-EPOCH was favored significantly over R-CHOP for OS (P < .01) (Fig. 1a) and EFS (P < .001) (Fig. 1b). When stratified by aaIPI, similar benefits were observed for R-EPOCH in patients with low aaIPI risk (Fig. 1c,d) and high aaIPI risk (Fig. 1e,f).

Figure 1.

These Kaplan-Meyer survivals illustrate (a,c,e) event-free survival and (b,d,f) overall survival for patients who received rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) or rituximab plus infusional prednisone, vincristine, cyclophosphamide, and doxorubicin (R-EPOCH) for (a,b) all patients and (c-f) when stratified by age-adjusted International Prognostic Index (IPI) risk group.

Relation Between Treatment and Clinical Outcomes After Adjustment for Baseline Covariates by Multivariate Analysis

Results of the multivariate analyses are provided in Table 3. The OR for achieving a CR was significantly greater in patients with low-risk aaIPI scores (OR, 4.6; P < .001) and a CD4 count ≥100/μL (OR, 2.7; P = 01). There was a trend toward an improved CR rate for R-EPOCH that was not statistically significant (OR, 1.9; P = .12). With regard to EFS, the features associated with a significantly lower HR for 2-year EFS included both low risk aaIPI (HR, 0.30; P < .001) and a CD4 count ≥100/μL (HR, 0.42; P < .001). Low-risk aaIPI and a high CD4 count also were associated with significantly improved 2-year OS rates. When adjusted for these covariates, treatment with R-EPOCH was associated with significant improvements in the EFS rate (HR, 0.4; P < .001) and the OS rate (HR, 0.38; P < .001). In other words, patients who received R-EPOCH had 60% lower likelihood of death or progressive lymphoma and 62% lower risk of death compared with patients who received R-CHOP after adjusting for the baseline covariates aaIPI and CD4 count. When the analysis was restricted to patients with DLBCL, similar benefits with respect to CR, EFS, and OS were observed (Table 4).

Table 3. Multivariate Analysis Assessing the Outcomes Complete Response, Event-Free Survival, and Overall Survival Adjusted for the Other Prognostic Covariates
 OR (95% CI)HR (95% CI)
  1. Abbreviations: OR, odds ratio; HR, hazard ratio; CI, confidence interval; CR, complete response; EFS, event-free survival; IPI, International Prognostic Index; R-CHOP, rituximab plus cyclophosphamide, vincristine, doxorubicin, and prednisone; R-EPOCH, rituximab plus etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin; OS, overall survival.

Age-adjusted IPI: Low risk vs high risk4.60 (1.97-10.72)0.32 (0.17-0.57)0.28 (0.14-0.55)
Baseline CD4 count: ≥100/μL vs <100/μL2.69 (1.25-5.76)0.42 (0.26-0.69)0.38 (0.22-0.64)
Treatment: R-EPOCH vs R-CHOP1.92 (0.86-4.22)0.40 (0.23-.69)0.37 (0.20-0.68)
Table 4. Outcomes of Patients With Diffuse Large B-Cell Lymphoma Treated With Concurrent Rituximab Plus Etoposide, Prednisone, Vincristine, Cyclophosphamide, and Doxorubicin (R-EPOCH) (n = 35) Compared With Rituximab Plus Cyclophosphamide, Vincristine, Doxorubicin, and Prednisone (R-CHOP) (n =80)
 OR (95% CI)HR (95% CI)
  1. Abbreviations: CI, confidence interval; CR, complete response; EFS, event-free survival; HR, hazard ratio; OR, odds ratio; OS, overall survival.

All patients1.92 (0.86-4.25)0.40 (0.23-0.69)0.37 (0.20-0.68)


The rate of treatment-associated deaths in both trials is summarized in Table 5. The treatment-associated mortality rate was 14% for R-CHOP in AMC010 and 9% for R-EPOCH in AMC034.11, 19 Treatment-associated deaths occurred significantly more often in patients with CD4 counts <50/μL when the results from both trials were combined (37% vs 6%; P < .01) and individually for patients who received either R-CHOP (36% vs 6%, P < .001) or R-EPOCH (38% vs 5%; P = 02). The cause of treatment-related death for patients with CD4 counts <50/μL was cryptosporidiosis, sepsis, and John Cunningham (JC) virus infection in AMC034 (n = 3) and sepsis syndromes in AMC010 (n = 8).

Table 5. Risk of Treatment-Associated Death According to Baseline CD4 Count
 No./Total No. (%)
Baseline CD4 CountAMC Trial 010: R-CHOP, n = 99AMC Trial 034: R-EPOCH, n = 51
  • Abbreviations: AMC, AIDS (Acquired Immunodeficiency Syndrome) Malignancy Consortium; R-CHOP, therapy with rituximab, cyclophosphamide, vincristine, doxorubicin and prednisone; R-EPOCH, therapy with concurrent rituximab, etoposide, prednisone, vincristine cyclophosphamide, and doxorubicin; TRM, treatment related mortality.

  • a

    P values are from a comparison between patients with CD4 counts <50/μL vs ≥50/μL.

Total TRM: All patients13/99 (13)5/51 (9)
<50/μL8/22 (36)3/8 (38)
≥50/μL5/77 (6)2/43 (5)


In a pooled analysis of sequentially performed trials for HIV-associated, aggressive B-cell NHL, we observed improved clinical outcomes for patients who received rituximab plus infusional EPOCH compared with rituximab plus standard CHOP chemotherapy, including significantly improved EFS and OS. This benefit likewise was observed after adjustment for baseline covariates that were associated with inferior clinical outcomes, including at least 2 poor-risk IPI features (stage III-IV disease, elevated lactate dehydrogenase, or poor performance status) and a low CD4 count (<100/μL) at the time of lymphoma diagnosis. The studies were performed by the same clinical trials organization and group of clinical investigators. Antiretroviral therapy was at the discretion of the treating physician in AMC034 and was used concurrently with chemotherapy in 70% of the patients, whereas it was mandatory in AMC010. Antibacterial and antifungal prophylaxis was not required in AMC010 but was mandatory in AMC034. Other guidelines for supportive care were similar between the 2 studies. These similarities should mitigate the impact of unknown or unmeasured factors on the improved outcomes we observed. Moreover, in AMC034, the EPOCH arm followed sequentially by rituximab did not meet the prespecified efficacy endpoint, which was met for patients who were randomized to receive concurrent R-EPOCH, providing additional evidence that the benefits of concurrent R-EPOCH indeed were attributable to that specific therapeutic regimen. Similar benefits for R-EPOCH were observed in both the low-risk aaIPI group and the high-risk aaIPI group. We also observed that patients who received R-CHOP or R-EPOCH and had CD4 counts <50/μL experienced a 35% to 40% rate of treatment-related mortality, suggesting that alternative therapeutic approaches should be considered in this population. For patients with CD4 counts ≥50/μL, both regimens were associated with similar treatment-associated mortality rates of approximately 5% to 6%, indicating a more favorable therapeutic index for rituximab.

The efficacy of R-EPOCH observed in the multi-institutional AMC034 trial is also consistent with the results reported from other trials that were performed at individual centers. Little and colleagues21 demonstrated a CR rate of 74% with a progression-free survival (PFS) rate of 73% and an OS rate of 60% at 53 months in patients with HIV-associated NHL who received dose-adjusted EPOCH; the addition of rituximab improved these outcomes in a study by Dunleavy and colleagues,18, 22, 23 who reported a CR rate of 92% and 4-year PFS and OS rates of 86% and 70%, respectively, for patients who received EPOCH-RR (rituximab on days 1 and 5 of each cycle).

We adjusted our clinical outcome estimates for the 2 strongest predictors of outcome in HIV-associated lymphomas: IPI score and CD4 cell count. The aaIPI is a validated and commonly used index for predicting outcomes in immunocompetent patients with aggressive NHL,24 and it has also been validated in patients with HIV-associated NHL.25-28 Low CD4 counts consistently have been associated with poor outcomes in the pre-cART and cART era.5, 27-29 When considered together in the same patient population, Bower and colleagues observed that the IPI and the CD4 cell count remained the only independent predictors for mortality in patients diagnosed with HIV-associated NHL in the cART era,30 supporting our strategy of including only these 2 baseline covariates in addition to treatment in our analysis.

The main limitations of our analysis are that patients were not concurrently randomized to R-CHOP versus R-EPOCH, and the analysis was conducted in a post-hoc manner. Although we controlled for the most important prognostic factors (CD4 count and aaIPI score), the association of R-EPOCH with improved clinical outcomes compared with R-CHOP still may be secondary to unaccounted confounding factors.

In summary, our finding that the treatment of HIV-associated lymphomas with R-EPOCH results in superior outcomes compared with R-CHOP is consistent with the results from other trials assessing the efficacy of these regimens. These results provide level 2 evidence supporting the use of R-EPOCH for selected patients with HIV-associated B-cell NHL and a CD4 count ≥50/μL. Benefits were observed in both high-risk and low risk patients but were more pronounced in the high-risk IPI group. Based on these findings, the ongoing AMC075 trial includes R-EPOCH as the control arm for patients with HIV-associated DLBCL and high aaIPI risk or other poor risk features (ie, high Ki67 proliferation rate) by comparing R-EPOCH alone or in combination with the histone deacetylase inhibitor vorinostat. However, additional confirmation is required in prospective, randomized clinical trials. Currently, this is the objective of an ongoing effort led by the Cancer and Leukemia Group B (CALGB), in which immunocompetent patients with DLBCL are randomized in a 1:1 fashion to receive either R-CHOP or R-EPOCH (CALGB 50303; national clinical trials no. NCT0118209). The results of this trial eventually will provide level 1 evidence that may define a role for infusional R-EPOCH therapy in DLBCL.


This work was supported an American Society of Clinical Oncology (ASCO) Cancer Foundation 2010 Young Investigator Award; by Clinical and Translational Science Awards (CTSA) grants UL1 RR025750, KL2 RR025749, and TL1 RR025748 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH); by the NIH Roadmap for Medical Research; and by AIDS Malignancies Consortium grant U01CA121947. The contents are solely the responsibility of the authors and do not necessary represent the official view of the NCRR, NIH, or ASCO.


The authors made no disclosures.