Changing incidence and prognostic factors of survival in AIDS-related ...

Leukemia & Lymphoma, February 2005; 46(2): 207 – 215

Changing incidence and prognostic factors of survival in AIDS-related non-Hodgkin’s lymphoma in the era of highly active antiretroviral therapy (HAART)

TIMO WOLF1, HANS-REINHARD BRODT1, STEPHAN FICHTLSCHERER2, KATHLEEN MANTZSCH1, DIETER HOELZER1, EILKE B. HELM1, PARIS S. MITROU1, & KAI UWE CHOW1 1

Department of Internal Medicine III, Hematology, Oncology and Infectious Diseases, University Hospital Frankfurt, Germany, and 2Department of Internal Medicine IV, Cardiology, University Hospital Frankfurt, Germany

Abstract Non-Hodgkin’s lymphoma is an AIDS-defining disease. The impact of HAART on the epidemiology and prognosis is debated controversially. A retrospective analysis has been performed in order to determine the influence of HAART. We collected data of 214 cases of AIDS-related Lymphoma (ARL) treated at our centre from January 1984 until May 2003 and analysed them using the Kaplan-Meier-, log rank- and Cox proportional hazard-model. The incidence of ARL increased between 1991 and 1994 up to a peak of 14.83 per 1000 patient years. In the subsequent periods from 1995 onwards however, it decreased to 3.7 in 1000 patient years. The incidence of AIDS-related primary CNS lymphomas (PCNSL) took a comparable, yet more pronounced development. Using the univariate Kaplan-Meier analysis prolonged survival was significantly associated with the achievement of a complete remission as well as with a favourable virological response to HAART. No significant differences could be shown for the use of protease inhibitors as well as for virological response being achieved before the diagnosis of NHL. When using the Cox model, complete remission overrides viral response and thus remained the only independent prognostic factor. Classical prognostic factors (CD4 count, prior Kaposi Sarcoma, extranodal manifestation, staging and histological subtype of NHL) were no longer significant for HAART patients in the multivariate analysis. These results illustrate the requirement for new prospective studies in order to determine the best options and ideal timing of coadministering chemotherapy and the type of HAART. Furthermore this study demonstrates that HAART decreases the incidence of ARL, and that achievement of a complete remission in patients suffering from ARL is—according to the multivariate analysis—the single most important prognostically relevant factor with respect to the time of survival.

Keywords: NHL, HAART, HIV, CR

Introduction Non-Hodgkin’s lymphoma (NHL) is an AIDSdefining disease [1]. Patients with HIV-infection are at increased risk of developing NHL. These lymphomas are mostly high-grade B-cell NHLs, characterized by rapid disease progression, frequent extranodal manifestations and a poor outcome [2]. The impact of highly active antiretroviral therapy (HAART) on the incidence of AIDS-related lymphoma (ARL) remains controversial. Several studies described reduced or stable incidence rates for ARL up to the year 2000 [3 – 10]. In contrast to other

AIDS-defining diseases, the incidence rate decreased less dramatically. Compared to the median survival of approximately 7 months in the pre-HAART era [11,12], the survival of ARL patients is prolonged in the post-HAART period [13 – 15]. This improvement was shown to be associated with higher rates of complete remission and response to HAART [14,15]. The standard treatment of ARL was and remains combination chemotherapy. Neither intensive chemotherapy regimens nor new combinations of chemotherapeutic agents have resulted in a better outcome as achieved by standard chemotherapy

Correspondence: Kai Uwe Chow, University Hospital Frankfurt, Department of Internal Medicine III, Hematology and Oncology, Theodor-Stern-Kai 7, 60590 Frankfurt am Main, Germany. Fax: 49-69-7373. E-mail: [email protected] Received for publication 5 October 2004. ISSN 1042-8194 print/ISSN 1029-2403 online # 2005 Taylor & Francis Group Ltd DOI: 10.1080/10428190400015733

208

T. Wolf et al.

treatment with CHOP-like therapy [16]. A coadministration of HAART and chemotherapy is possible and results in better immune function [17]. The role of the monoclonal anti-CD20 antibody in the treatment of ARL is not yet defined. High rates of serious infectious complications were recently described [18]. AIDS-related lymphoma show two major histological categories, the small non-cleaved lymphoma, including the Burkitt-lymphoma, and the diffuse large cell lymphoma [2]. To clarify whether different prognostic factors including HAART may influence the incidence of ARL or the outcome of patients with ARL, we retrospectively analysed 214 cases of ARL from the HIV-cohort at the Frankfurt University Hospital, Germany, treated between January 1984 and May 2003. Methods The Frankfurt AIDS cohort study (FACS) is an ongoing, open, prospective cohort study of HIVinfected homosexual and bisexual men from the Frankfurt Rhein/Main area. Each patient receives follow-up evaluations including physical examinations, laboratory tests including CD4 cell counts and virus load at short intervals of 2 – 6 months. Data on clinical AIDS events and antiretroviral therapy are collected with standardized methods as described elsewhere [19]. For the present study, a retrospective cohort was established on the basis of the FACS. In addition HIV-infected patients with other modes of infection as the participants of the FACS were included, i.e. patients with intravenous drug use (IVDU), heterosexual contacts and those receiving blood products. These patients were examined and treated by the same physicians and received the same follow-up evaluations as mentioned above. ARL were classified according to the WHO-classification. Patients were staged according to the Ann Arbor classification for lymphoma. Diagnosis was confirmed by bone marrow- or lymph node- or extranodal tissue histology and histochemistry. CNS analysis was routinely performed in every patient during primary staging. Chemotherapy regimens including cyclophosphamide, doxorubicine, vincristine and prednisone (CHOP) or similar regimens (CHOP-like) were considered to be curative either with or without the use of monoclonal anti-CD20 antibody rituximab or radiation therapy. HAART was defined as a combination of at least three antiretroviral drugs. Viral response to HAART was defined as a viral load 5 500 copies/ml within 3 months after the begin of antiretroviral therapy (intent-to-treat). When patients experienced a viral rebound, this was judged as treatment failure.

Prognostic factors included in the analysis were sex, median age, B-symptoms, localization as well as histology of NHL, LDH, median CD4 count and median viral load. Statistics Survival curves were estimated by the method of Kaplan and Meier. Univariate analyses were performed using the Log rank test (Graph Pad Prism Software, USA), the Wilcoxon test were used for continuous, and the Likelihood ratio method for ordinal data. In case of small sample sizes, Fisher’s exact test was applied. In addition, Cox regression analysis was used to examine the potential relationships between continuous variables and events during the follow-up period. Multivariate analysis using Cox regression techniques was performed to examine potential interactions among the entered covariates. Statistical significance was assumed if the null hypothesis could be rejected at the P = 0.05 level. All statistical analysis was performed using SPSS for Windows 11.5 (SPSS Inc). Incidence was calculated based on 1000 patient years in five 4-year periods for the time from January 1983 until December 2002. Out of 214 patients, 211 were included in this analysis. Three patients of our cohort were diagnosed in 2003. The time periods were chosen according to fundamental changes in the availability and use of antiretroviral mono- and combination therapy. The follow-up time was calculated from the first presentation to our department until the earliest of either last follow-up, death or diagnosis of NHL according the methods described by Rickerts et al. [20]. Briefly, if a patient’s follow-up exceeded a four year period, the follow-up time was terminated with the end of the period and the remainder attributed to the subsequent period.

Results Decreasing incidence of ARL as well as of primary CNS lymphoma (PCNSL) in the era of HAART Figure 1 shows the incidence rates for ARL as well as PCNSL. Generally, the absolute number of patients treated in our cohort increased steadily up to 3384 patients between 1999 – 2002. Despite the increasing numbers of patients, the mean time of follow-up per patient was stable between 1987 and 1998 (2.05 – 2.145 years). In the last period however, it rose to 2.796 years, thus explaining the increase in cumulative patient years (n = 9463 years). Over the first three periods, relative numbers of ARL have increased steadily from 8.24 per 1000 years between 1983 and

HIV-associated NHL in the HAART era

209

Figure 1. Incidence of ARL and AIDS-related PCNSL at the Frankfurt University Hospital from 1983 – 2002. Absolute number of patients, cumulative patient years, average follow-up per patient are shown in the first three columns. Absolute and relative incidences are shown in table and graph for (systemic) ARL as well as PCNSL.

1986, to 14.83 per 1000 years between 1991 and 1994, the period with the highest incidence for both NHL and PCNSL. As protease inhibitors became available in 1995, first in clinical studies, later for all patients, this development has been reciprocated. The incidence of NHL decreased to 8 per 1000 patient years between 1995 and 1998 and more recently to 3.7 per 1000 patient years (1998 – 2002). The same could be found for the absolute numbers of PCNSL in the respective periods, as shown in Figure 1. Both curves are running concurrently. These data show that through the use of HAART, the morbidity of ARL has been reduced. Baseline characteristics We followed a total of 214 patients in this study, the baseline characteristics of them are shown in Table I. For 61% (131) of our patients, NHL was the first AIDS-defining event, while 10.7% (23) had Kaposi’s Sarcoma before NHL could be diagnosed. Table I compares patients’ baseline values from pre-HAART and post-HAART periods. As expected, the number of patients who had AIDSdefining diseases before the diagnosis of NHL was significantly lower in the HAART-era. Likewise, the

percentage of patients who had NHL as their AIDSdefining diagnosis in the HAART-era was larger than before. As the spectrum of opportunistic diseases has changed with the introduction of effective antiviral therapy, other AIDS-defining illnesses as primary manifestations have declined and an increasing number of our patients (74,1%) had NHL as the primary AIDS-defining diagnosis. Patients were often found to have manifest immunosuppression at the time of primary diagnosis of NHL. The median CD4 count of all our patients was 93 ml and 53% of all patients had less than 100 ml. As can be expected, these figures were significantly improved after highly active antiretroviral therapy had been introduced. In the HAART period, the median CD4 count was 102 ml and 39% of patients had less than 100 CD4 cells ml. In the pre-HAART period in contrast, the median CD4 count was at 66 ml and 74% of patients had less than 100 ml. All other characteristics showed no significant differences between HAART and preHAART patients. Extranodal manifestations were frequent, since 66% of all patients had an involvement of at least one extranodal site. Meningeal involvement was found in 11%. Patients mostly presented with advanced NHL disease. Thus 47% were found to

210

T. Wolf et al.

Table I. Baseline characteristics and subdivision in pre- and post-HAART patients. all (%) Whole population Male/female Age (years) median Age (years) range

214 (100) 199/15 38 21,7 – 71

pre-HAART era (%) 129 (100) 121/8 (93.7/6.3) 36.7 21,7 – 64

post-HAART era (%) 85 (100) 78/7 (91.8/8.2) 39.4 23,4 – 71

P

0.0188

Risk MSM Hetero IVDA

141 (66.0) 18 (8.4) 55 (25.6)

90 (69.9) 13 (9.9) 26 (20.2)

51 (60.0) 5 (5.8) 29 (34.2)

NHL as primary AIDS manifestation

131 (61.2)

68 (52.7)

63 (74.1)

0.0126

83 (38.8) 23 (10.7) 60 (28.1)

61 (47.3) 18 (14.0) 43 (33.3)

22 (25.9) 5 (5.9) 17 (20)

0.0213 0.0473

B-symptoms A Unknown

105 (49) 54 (25.2) 55 (25.8)

54 (41.9) 30 (23.3) 45 (34.8)

51 (60) 24 (28.2) 10 (11.8)

Localization of NHL At least one extranodal manifestation Meningeal involvement Strictly nodal manifestation Unknown

141 24 51 22

(65.9) (11.2) (23.8) (10.3)

88 18 24 17

(68.2) (14.0) (18.6) (13.2)

53 6 27 5

(62.4) (7.1) (31.8) (5.8)

Ann Arbor Stage I II III IV Unknown

33 33 24 102 22

(15.4) (15.4) (11.2) (47.7) (10.3)

20 21 13 58 17

(15.5) (16.3) (10.1) (45.0) (13.1)

13 12 11 44 5

(15.3) (14.1) (12.9) (51.8) (5.9)

NHL Histology T-NHL DLC Burkitt

10 (4.7) 162 (75.7) 42 (19.6)

5 (3.9) 93 (72.1) 31 (24.0)

CD4 counts median CD4 counts range Absolute CD4 cell count 5 100

93 (0 – 1228) 113 (52.8)

66

Other primary AIDS manifestation KS Others

102

(0 – 1028) 74 (57.4)

Viral load, log copies median Range Viral load below 500 cop/ml

5 (5.9) 69 (81.2) 11 (12.9)

n.d. n.d. n.d.

0.0136

(1 – 1228) 39 (45.9) 66500 (4 500 – 5,000,000) 16 (18.8)

P-values are shown in case of significant differences between the groups.

be in Ann Arbor stage IV at the initial presentation and 49% exhibited B-symptoms. According to the REAL classification, 75.7% of patients had diffuse large-cell lymphoma (DLC), 19.6% a Burkitt-like histology and 4.7% high-grade T-cell lymphoma. Interestingly, the ratio of histological types of ARL has shifted between the two respective eras. Burkittlike lymphoma was found in 24% of cases before and 12.9% after the introduction of HAART. Correspondingly, the ratio of DLC histology was increased while the one for lymphomas of T-cell origin remained stable (Table I). This however is an

observation made on very few cases, and the findings were not statistically significant. Chemotherapy and antiretroviral therapy We next examined the subgroup of patients that had received HAART. Among the 85 patients from the HAART period, 61 (72%) had actually received HAART. The remaining 24 had either received a non-HAART combination of antiretrovirals or no therapy at all, due to late presentation to our department in a very advanced disease of NHL,


211

intent. This difference was reflected in the achieved rates of complete remission. After HAART, 28 patients achieved a complete remission (CR), as compared to only 5 patients in the pre-HAART period. This means that 38% of patients with a curative regimen achieved CR after 1995, whereas only 8.5% before the HAART-era had this outcome. Amongst the group of patients who had responded to HAART, the proportion of Burkitt’s lymphoma was larger (16.7%) than among the nonresponders (9.7%). This finding was not statistically significant.

non-compliance or refusal. Table II summarizes the outcome and baseline characteristics for HAART patients according to the response to antiviral therapy. About half (n = 30) of the HAART receivers had responded virologically. As shown in Table II, 40 patients (47%) had been receiving HAART before they became diagnosed with lymphoma, and 14 of them developed NHL with a virus load of less than 500 ml, as they had responded to HAART (Table III). Concerning the chemotherapy regimen, 85.9% of HAART patients and only 52.7% of pre-HAART patients received chemotherapy with a curative

Table II. Baseline characteristics of antiretroviral therapy. Results at the time of NHL diagnosis are shown; HAART responders and nonresponders are displayed in comparison. Antiretroviral Therapy post-HAART era (%) Patients in HAART era

85 (100)

non-HAART therapy HAART receiver

24 (28.2) 61 (71.8)

HAART response HAART no response

30 (35.3) 31 (36.5)

HAART before NHL HAART response before NHL HAART no response before NHL

40 (47.1) 14 (16.5) 26 (30.6)

(4 patients had response after NHL)

HAART after NHL HAART response after NHL HAART no response after NHL

Whole population Male/female Age, years (median)

21 (24.7) 12 (14.1) 9 (10.6) HAART-responders HAART non-responders 30 (100%) 31 (100%) 27/3 (90%/10%) 28/2 (93.3%/6.7%) 40.6 (23.4 – 71.0) 39,1 (30.6 – 60.7%)

all 61 55/5 39.8 (23.4 – 71.0)

B-symptoms A Unknown

15 (50%) 14 (46.7%) 1 (3.3%)

21 (67.7%) 6 (19.4%) 4 (12.9%)

36 (59%) 20 (33%) 5 (8.2%)

Localization of NHL At least one extranodal manifestation Meningeal involvement Strictly nodal manifestation Unknown

18 (60%) 1 (3.3%) 12 (40%) 0

21 (67.7%) 0 7 (22.6%) 3 (9.7%)

39 1 19 3

NHL Histology T-NHL DLC Burkitt

0 25 (83.3%) 5 (16.7%)

3 (9.7%) 25 (80.6%) 3 (9.7%)

3 (4.9%) 50 (82.0%) 8 (13.1%)

LDH (median)

220 (125 – 1503)

CD4 cell counts median Absolute CD4 cell count 5 100

184 (8 – 1228) 10 (3.3%)

Viral load, log copies median

275

P-values are not shown when not significant.

261 (147 – 1770) 83 (2 – 685) 17 (56.7%) 186000

P

(64%) (1.6%) (31.1%) (4.9%)

239 (125 – 1770) 102 (1 – 1228) 27 (44.3) 93126

0.0231

not done

212

T. Wolf et al.

Table III. Characteristics of chemotherapies. Chemotherapy All patients (%) Patients Curative chemotherapy

214 (100) 141 (65.9)

CHOP/CHOP like CHOP + Rituximab CHOP + Radiatio + Rituximab CHOP/CHOP like + Radiatio

114 6 1 20

Others (only Radiation or Surgery) None Unk CR No CR Unknown

Pre-HAART era (%) 129 (100) 68 (52.7)

Post-HAART era (%) 85 (100) 73 (85.9)

(53.3) (2.8) (0.5) (9.3)

59 (45.7) 0 0 9 (7.0)

5 (2.3) 6 (2.8) 62 (29.0)

4 (3.1) 2 (1.6) 55 (42.6)

1 (1.2) 4 (4.7) 7 (8.2)

33 (15.4) 148 (69.2) 33 (15.4)

5 (3.9) 99 (76.7) 25 (19.4)

28 (32.9) 49 (57.6) 8 (9.5)

In the univariate analysis HAART-response and achievement of a complete remission (CR) are associated with improved survival The survival analyses for our cohort are shown in Figures 2 and 3. When comparing patients from the pre-HAART (n = 129) to those from the HAARTperiod (n = 85), the latter exhibited a significantly better survival. The median survival for HAART patients was 276.0 days as compared to 135.0 days (Figure 2A). Out of the 85 post-HAART patients, 61 had actually received HAART. Within the group of HAART receivers, it is interesting to see whether an actual virological response to HAART (n = 30) confers an advantage as compared to the cases where patients received HAART but failed virologically (n = 31). Survival for HAART responders was significantly better. Non-responders had a median survival period of 201 days, whereas the median survival was not yet reached in the responder group (Figure 2B). In the next figure, the survival is shown only for those patients who were treated with HAART at the time the NHL was diagnosed (n = 40), and had either a virological suppression before (n = 14) the diagnosis of lymphoma or not (n = 26). Although the patients exerting virological control before ARL showed a better survival curve, and although they had not reached 50% survival against a median survival of 178 days for those who responded after the diagnosis or ARL, these differences were not significant so far (P = 0.0671, Figure 2C). In order to determine whether a difference in the applied HAART-regimen is relevant for the overall prognosis, we examined PIcontaining and PI-sparing therapies. However, no significant difference could be found (Figure 3A).

55 6 1 11

(64.7) (7.1) (1.2) (12.9)

Next, Kaplan-Meier curves show patients with and without achieved complete remission after chemotherapy. The data of the 85 patients in the HAART era and of the 61 who had actually received HAART were analysed separately (Figure 3B and data not shown) for the outcome following chemotherapy. Analysing all patients in the HAART-era (n = 85) demonstrates that survival was significantly better if patients went into CR (n = 28). Thus the median survival in this subpopulation has not been reached, whereas patients (n = 49) who did not achieve a CR had a median survival of only 149 days, in 8 patients the remission status was not determined (Figure 3B and Table II). Similar results were achieved calculating Kaplan-Meier plots for the subgroup of the 61 HAART-receivers (data not shown). Except the expected differences in immunological and virological data, the baseline characteristics of the HAART-responder/non-responder as well as the CR/no CR subgroups were similar concerning age, gender and NHL stage (data not shown). In the further univariate Kaplan-Meier analysis classical prognostic factors, such as lymphoma stages I/II vs. III/IV (P = 0.0092), CD4 counts 4 200 ml vs. 5 200/ml (P = 0.005) as well as the absence of B-symptoms vs. presence of B-symptoms (P = 0.0016) were associated with an improved OS. No association between prolonged and the baseline LDH level, extranodal involvement, lymphoma subtype or age could be found (data not shown). However, some of these prognostic factors were included into the multivariate model, as they had been shown to be strong predictors of survival in previous studies.

213


The prognostic value of a documented CR is superior to the HAART-response in the multivariate analysis Using the Cox proportional hazard model, the subgroup of 61 HAART receivers was analysed. Only the achievement of CR functions as an independent predictor of survival with a hazard ratio of 3.197 (P = 0.006). In contrast, response to HAART was not significant in the multivariate analysis as reflected by the large confidence interval (Table IV). All other factors included in the multivariate model even had a lower level of statistical significance, including the prognostic factors concerning the stage and histology of NHL, the prior existence of KS, primarily extranodal manifestation and CD4 count. Discussion

Figure 2. Kaplan-Meier analysis of (A) survival in pre- and postHAART era, (B) survival in HAART-responders or non-responders and (C) patients who achieved viral response before or after NHL was diagnosed.

The incidence of ARL has only recently been thought to decrease. Studies that covered time periods until the end of 1997 showed stable or only slightly incidence rates [7,8]. Two multi-institutional studies showed a reduction in incidence rates from 1996 and 1997 onwards [9,10]. We are an institution which was able to administer HAART comparatively early. In our large, unicentric cohort, the follow-up period after the introduction of HAART is longer than in previous studies, and data have been collected until 2002. Our results show that the incidence of NHL has been clearly reduced since 1995. This was the time in which protease inhibitors and hence highly active antiretroviral therapy became available. In our methodology, we used 4 year-periods that corresponded to the respective developmental stage of ART and the changing incidence corresponded to the introduction of proteases inhibitors and thus HAART as such. The decrease in incidence can be explained by several factors. One is the accumulation of follow-up

Table IV. Results of the multivariate analysis of independent predictors of survival, cox-regression model and hazard ratio in 61 patients receiving HAART. Factor No prior KS Stage (4 II) Burkitt Strictly nodal CD4 count a) Viral response CR of NHL

Figure 3. Kaplan-Meier analysis of (A) subgroups of HAART patients with use of no, one or two protease inhibitors, (B) subgroups of patients with or without complete remission.

HR

95% CI

P

4.225 0.774 0.774 0.792 0.656 1.904 3.197

0.505 – 35.333 0.823 – 2.438 0.169 – 3.547 0.076 – 29.018 0.268 – 1.606 0.100 – 36.271 1.403 – 7.281

0.184 0.209 0.741 0.792 0.356 0.668 0.006

(HR 4 1 indicates improved survival; a) CD4 counts in tertials 4 180, 180 – 67, 5 67). Factors not significant in the univariate analysis were not included in the multivariate model, except for CD4-count.

214

T. Wolf et al.

time per patient expressed in the increase in patient years in our cohort, due to a better survival. We would however like to stress that the absolute cases of NHL have also decreased clearly. This reduction is running in parallel with a more pronounced reduction of incidence in PCNSL, as reported previously [13]. The interpretation of these findings would be that immune function is improved and NHL subsequently occurs less frequently, as their natural history could be influenced by the immune reconstitution. However, there are no data at hand to prove this hypothesis, and it is beyond the limitations of our data set to show reliably that the control of HI Viremia reduces the incidence of ARL. The fact that immunological and virological parameters at baseline have improved due to HAART is undebated and has been demonstrated in our cohort. Interestingly, within the HAART period, the proportion of Burkitt(-like) types appears to be reduced, whereas the ratio of DLC histology appears to be increased compared to pre-HAART, although there was no statistical significance. This development is discrepant to results of other groups, who have shown that the reduction in incidence of DLC is much more pronounced as for Burkitt’s lymphoma. We would like to stress that the opposite observation was made in our cohort when comparing those patients who actually responded to HAART to those who did not. It is thinkable that this development of incidences is not yet reflected in the relative prevalence of the complete post-HAART-era. Overall, these data suggest that the prognosis of Burkitt’s NHL is not improved as strongly as seems to be the case in other subtypes. Whether or not there is a link between improved immunological function and the occurrence of different NHL subtypes is well beyond the limitations of this study, but is an interesting question to be addressed in future studies. Other groups have shown a number of prognostic factors for populations that included HAART-users and non-users alike [21], such as NHL characteristics, CD4 count and prior AIDS-defining diseases. As more and more patients do develop ARL while being on HAART nowadays, we performed a subanalysis on HAART users. Other prognostic factors that were considered to be important previously, like the initial stage and histology of NHL, extranodal manifestation and baseline LDH and CD4 count did no longer show a significant association with survival in our study when looking at HAART-users only. We would like to suggest that after the introduction of HAART, effective chemo- and antiretroviral therapy are the single most significant predictors of survival, and ‘‘classic’’ prognostic factors are of less value. Survival analysis showed that achieving a complete remission is a strong independent predictor of overall

survival. This finding is well in agreement with the results which other groups have shown [15]. Although viral response was associated with prolonged survival in the univariate analysis, we found that complete remission is a stronger predictor than viral response in the multivariate model. In practice, it can be difficult to coadminister chemotherapy and anti-retroviral therapy. The dilemma for the physician is often that one type of therapy needs to be delayed in favour of the other. These findings suggest that effective chemotherapy should be the first priority, but HAART should be started as early as possible. It should be mentioned that the remission rate in our cohort before the introduction of HAART was less than 10%. This is very low, lower indeed than what other groups have shown [13,23]. It is not easy to explain this discrepancy, but many of the other studies were conducted prospectively. This work is a retrospective, single institution analysis. Concerning the role of HAART, we could show that survival was significantly better for those who had successful HAART, as others have also published [15,21]. Additionally, results have been published showing that concomitant HAART and chemotherapy does not have a stronger burden of resistance development. When trying to find an optimal HAART for a patient with ARL, there are a number of important aspects to take into account. First of all, we were trying to see whether effective HAART before the diagnosis and treatment of lymphoma improves survival. Although no statistical difference was found, survival curves were better for patients with virological suppression before lymphoma. This seems to indicate that it is beneficial to initiate HAART early. Another study has shown that the outcome is improved by long-term HAART use ( 4 24 months) before diagnosis and treatment of NHL [21]. This may suggest that it is still too early to measure a strong statistical benefit concerning the overall survival for virological suppression. We also attempted to examine other aspects of HAART, such as PI use, but our case numbers were too low to thoroughly address this problem. Although HAART has reduced the incidence of lymphoma, there are a number of issues to address as far as the treatment of ARL is concerned. There is an urgent need in our view to address the optimal cotherapy in the framework of a study that goes deeper than the sheer feasibility of coadministering HAART and chemotherapy. The best used drug combinations and optimal timing in administration will have to be defined. Although the incidence of ARL has decreased, we will be required to optimize chemotherapy and HAART to be able to combine these in the best possible way in order to deliver adequate medical care.

HIV-associated NHL in the HAART era Acknowledgements We thank Prof. Schlomo Staszewski and Beverley Jennings for kindly providing data from their HIVdatabase.

References [1] Goedert JJ, Cote TR, Virgo P, Scoppa SM, Kingma DW, Gail MH, Jaffe ES, Biggar RJ. Spectrum of AIDS-associated malignant disorders [see comments]. Lancet 1998;351: 1833 – 1839. [2] Tirelli U, Spina M, Gaidano G, Vaccer E, Franceschi S, Carbone A. Epidemiological, biological and clinical features of HIV-related lymphomas in the era of highly active antiretroviral therapy. Aids 2000;14:1675 – 1688. [3] Grulich AE. AIDS-associated non-Hodgkin’s lymphoma in the era of highly active antiretroviral therapy. J Acquir Immune Defic Syndr 1999;21(Suppl 1):S27 – S30. [4] Jacobson LP, Yamashita TE, Detels R, Margolick JB, Chmiel JS, Kingsley LA, Melnick S, Munoz A. Impact of potent antiretroviral therapy on the incidence of Kaposi’s sarcoma and non-Hodgkin’s lymphomas among HIV-1-infected individuals. Multicenter AIDS Cohort Study. J Acquir Immune Defic Syndr 1999;21(Suppl 1):S34 – S41. [5] Rabkin CS. Epidemiology of B-cell lymphomas. Curr Top Microbiol Immunol 1999;246:235 – 239, discussion 240. [6] Buchbinder SP, Holmberg SD, Scheer S, Colfax G, Om P, Vittinghoff E. Combination antiretroviral therapy and incidence of AIDS-related malignancies. J Acquir Immune Defic Syndr 1999;21(Suppl 1):S23 – S26. [7] Gerard L, Galicier L, Maillard A, Boulanger E, Quint L, Matheron S, Cardon B, Meignin V, Oksenhendler E. Systemic non-Hodgkin lymphoma in HIV-infected patients with effective suppression of HIV replication: persistent occurrence but improved survival. J Acquir Immune Defic Syndr 2002;30:478 – 484. [8] Carrieri MP, Pradier C, Piselli P, Piche M, Rosenthal E, Heudier P, Durant J, Serraino D. Reduced incidence of Kaposi’s sarcoma and of systemic non-Hodgkin’s lymphoma in HIV-infected individuals treated with highly active antiretroviral therapy. Int J Cancer 2003;103:142 – 144. [9] Kirk O, Pedersen C, Cozzi-Lepri A, Antunes F, Miller V, Gatell JM, Katlama C, Lazzarin A, Skinhoj P, Barton SE. Non-Hodgkin lymphoma in HIV-infected patients in the era of highly active antiretroviral therapy. Blood 2001;98:3406 – 3412. [10] International Collaboration on HIV and Cancer. Highly active antiretroviral therapy and incidence of cancer in human immunodeficiency virus-infected adults. J Natl Cancer Inst 2000;92:1823 – 1830.

215

[11] Klencke B, Kaplan L. Advances and future challenges in non-Hodgkin’s lymphoma. Curr Opin Oncol 1998;10:422 – 427. [12] Straus DJ. HIV-associated lymphomas. Curr Opin Oncol 1997;9:450 – 454. [13] Chow KU, Mitrou PS, Geduldig K, Helm EB, Hoelzer D, Brodt HR. Changing incidence and survival in patients with AIDS-related non-Hodgkin’s lymphoma in the era of highly active antiretroviral therapy (HAART). Leuk Lymphoma 2001;41(1 – 2):106 – 116. [14] Gerard L, Galicier L, Boulanger E, Quint L, Lebrette MG, Mortier E, Meignin V, Oksenhendler E. Improved survival in HIV-related Hodgkin’s lymphoma since the introduction of highly active antiretroviral therapy. Aids 2003;17:81 – 87. [15] Hoffmann C, Wolf E, Fatkenheuer G, Buhk T, Stoehr A, Plettenberg A, Stellbrink HJ, Jaeger H, Siebert U, Horst HA. Response to highly active antiretroviral therapy strongly predicts outcome in patients with AIDS-related lymphoma. Aids 2003;17:1521 – 1529. [16] Kaplan LD, Straus DJ, Testa MA, Von Roenn J, Dezube BJ, Cooley TP, Herndier B, Northfelt DW, Huang J, Tulpule A, Levine AM. Low-dose compared with standard-dose mBACOD chemotherapy for non-Hodgkin’s lymphoma associated with human immunodeficiency virus infection. National Institute of Allergy and Infectious Diseases AIDS Clinical Trials Group [see comments]. N Engl J Med 1997; 336:1641 – 1648. [17] Powles T, Imami N, Nelson M, Gazzard BG, Bower M. Effects of combination chemotherapy and highly active antiretroviral therapy on immune parameters in HIV-1 associated lymphoma. Aids 2002;16:531 – 536. [18] Kaplan LD, Lee J, Scadden DT. No benefit from rituximab in a randomized phase iii trial of chop with or without rituximab for patients with HIV-associated non-Hodgkin’s lymphoma: updated data from aids malignancies consortium study 010. Blood 2003;102:409a. [19] Brodt HR, Kamps BS, Gute P, Knupp B, Staszewski S, Helm EB. Changing incidence of AIDS-defining illnesses in the era of antiretroviral combination therapy [see comments]. Aids 1997;11:1731 – 1738. [20] Rickerts V, Brodt H, Staszewski S, Stille W. Incidence of myocardial infarctions in HIV-infected patients between 1983 and 1998:the Frankfurt HIV-cohort study. Eur J Med Res 2000;5:329 – 333. [21] Vaccher E, Spina M, Talamini R, Zanetti M, di Gennaro G, Nasti G, Tavio M, Bernardi D, Simonelli C, Tirelli U, et al. Impact of concomitant antiblastic chemotherapy and highly active antiretroviral therapy on human immunodeficiency virus (HIV) viremia and genotyping in HIV-infected patients with non-Hodgkin lymphoma. Clin Infect Dis 2003;37: 1556 – 1564.