Characteristics of non-Hodgkin lymphoma arising in HIV-infected patients with suppressed HIV replication
13 November 2009
Gerard, Laurence; Meignin, Veronique; Galicier, Lionel; Fieschi, Claire; Leturque, Nicolas; Piketty, Christophe; Fonquernie, Laurent; Agbalika, Felix; Oksenhendler, Eric
aService d'Immunopathologie Clinique, Hopital Saint-Louis, Assistance Publique-Hopitaux de Paris, France
bLaboratoire de Pathologie, Hopital Saint-Louis, Assistance Publique-Hopitaux de Paris, INSERM U728, Universite D. Diderot Paris VII, France
cService d'Immunologie Clinique, Hopital Europeen Georges Pompidou, Assistance Publique-Hopitaux de Paris, France
dService de Maladies Infectieuses et Tropicales, Hopital Saint-Antoine, Assistance Publique-Hopitaux de Paris, France
eLaboratoire de Virologie, Hopital Saint-Louis, Assistance Publique-Hopitaux de Paris, Universite D. Diderot Paris VII, Paris, France.
Objective: Despite effective treatment of HIV infection, some patients still develop non-Hodgkin lymphoma (NHL). We analysed patients with HIV-associated NHL and undetectable plasma HIV-RNA, according to the duration of HIV suppression.
Methods: Out of 388 patients included in a prospective cohort of HIV-associated NHL from 1996 to 2008, 128 (33%) had a plasma HIV-RNA below 500 copies/ml and were included in the study. Patients with long-term HIV suppression (>18 months) were compared with patients with recent HIV suppression (≤18 months).
Results: All patients but three were treated with combination antiretroviral therapy, with a median duration of 2.2 years. The median duration of HIV suppression was 10.1 months. Most cases (65%) occurred within 18 months following HIV suppression. In the more than 18 months group, patients developed NHL at a higher CD4 cell count than patients with 18 months or less of HIV suppression (359 versus 270 cells/µl, P = 0.02). None of the NHL characteristics were different between the two groups. Outcome was similar in the two groups (complete remission, 64 versus 72.5%; P = 0.35 and 3-year survival, 46 versus 56%; P = 0.08). In addition, 52% of the tumours were Epstein-Barr virus or human herpesvirus 8 associated, without any difference in the proportion of virus-associated tumours according to the duration of HIV suppression.
Conclusion: In patients with undetectable HIV-RNA, NHL occurred mainly within the first 18 months following HIV suppression. In patients developing NHL after long-term HIV suppression, the level of CD4 cell count was higher, but the association with Epstein-Barr virus or human herpesvirus 8 and the prognosis were similar to that observed in patients with recent HIV suppression.
Since the introduction of effective combination antiretroviral therapy (cART), HIV-associated morbidity and mortality rates have dramatically fallen [1-3]. The incidence of HIV-associated non-Hodgkin lymphoma (HIV-NHL) has also decreased significantly, with the most dramatic decline observed in primary central nervous system lymphoma (PCNSL) [4-6]. However, this decline is less marked than in other HIV-associated complications after initiation of cART [4,7,8]. In HIV-infected patients, the relative risk for developing NHL remains around 13-20 in population-based studies, even in the late cART era [6,8,9]. In Western Europe, the proportion of AIDS-defining illness attributable to NHL has increased since the widespread use of cART from less than 4% in 1994 to almost 16% in 1998 . Although cART had a marked favourable impact on outcome [5,11-14], with complete remission rates close to that observed in patients with aggressive NHL in the general population, NHL remains a major cause of death in HIV-infected patients [15,16].
Characteristics of HIV-NHL have been extensively described in the pre-cART era. The clinical presentation is more aggressive, with a different spectrum of lymphoma pathological subtypes than in the general population. Up to 30% of HIV-NHL are considered to be Burkitt lymphoma, which is a rare B-cell proliferation in HIV-negative adults outside of endemic areas [13,14,17-19]. The association with some herpes viruses such as Epstein-Barr virus (EBV), detected in 25% of the Burkitt subtype and in 60-80% of the immunoblastic subtype of diffuse large B-cell lymphoma (IBL-DLBCL), and human herpesvirus 8 (HHV-8), detected in rare lymphoproliferative diseases, is mainly observed in the HIV setting [13,18,20-24]. Controversial data have been published on a possible change of HIV-NHL spectrum in the cART era [4,5,12,14,25].
Apart from CD4 cell count deficiency, recent studies [4,26,27] have demonstrated that incomplete viral suppression during cART was a strong, independent risk factor for the development of HIV-NHL. In France, about 80% of patients receiving cART have an undetectable plasma HIV-RNA (French Hospital Database on HIV, http://www.ccde.fr). However, despite effective ART and subsequent complete suppression of HIV replication, some patients still develop NHL. One could expect that, in patients successfully treated for HIV infection, the duration of the virological control could influence the clinical features and the pathological distribution of lymphomas arising in these patients, with a trend to move closer to those observed in the general population. The objective of this study was to determine, in a population of patients diagnosed with NHL while plasma HIV-RNA is undetectable, whether the characteristics of lymphoma vary with the duration of HIV suppression, and particularly, whether the association with EBV or HHV-8 decreases over time.
This prospective single-centre study, with extended follow-up in the cART era, demonstrated that one-third of HIV-NHL occurred in patients with complete suppression of HIV replication. Most of these lymphomas occurred early after HIV-RNA became undetectable (65% within 18 months). However, some patients still develop NHL after long-term HIV suppression. We have previously reported a short series of 27 patients from the same cohort in the early cART era and did not found any difference in the characteristics of patients with controlled HIV infection as compared with patients with untreated or treated but uncontrolled HIV infection . Herein, we extended the analysis to 128 patients, with a centralized review of all available pathological specimens by a single pathologist, associated with the detection of EBV and HHV-8 in the tumour cells. We assumed that NHL that occurred in patients with long-term suppression of HIV replication had characteristics closer to that of non-HIV-associated NHL, with less severe characteristics, decreased EBV/HHV-8 association, and improved outcome. This study did not support this hypothesis. As expected, duration of HIV infection and duration of cART were longer in patients with longer HIV suppression. However, patients with long duration of HIV suppression had similar lymphoma characteristics than that observed in patients with recent HIV suppression or persistent HIV replication.
In these patients successfully treated with cART, immune reconstitution was effective, with a higher CD4 cell count (297 cells/µl) at diagnosis of NHL than that observed in uncontrolled patients (155 cells/µl) from the same cohort and than that reported in the literature [4,11,12,17]. The difference between the CD4 cell count at diagnosis of NHL and the CD4 cell count nadir, indirectly reflecting immune reconstitution, was above 100 cells/µl in the majority of patients (63.5%) (data not shown). Increase in the CD4 cell count level was associated with duration of HIV suppression, and NHL occurred at a significantly higher CD4 cell count after longer duration of HIV suppression (359 cells/µl) than early after HIV suppression (270 cells/µl). The distribution of CD4 cell count by strata showed that only one patient developed NHL while having a CD4 cell count below 100 cells/µl after the first year of HIV suppression. This finding is consistent with previously published data, suggesting that the decreased incidence in NHL observed in cART era was explained by the decrease in the proportion of patients with low CD4 cell counts . Several studies [9,33] have shown an association between the CD4 cell count nadir and the development of NHL; indeed, in this study, almost all patients had a CD4 cell count nadir below 200 cells/µl (median 112 cells/µl). This could argue for an earlier introduction of cART, before the nadir CD4 cell count drops below a critical level.
Characteristics of NHL arising in patients with undetectable HIV-RNA remain different from that of lymphoma observed in the general population. Most cases are aggressive, with advanced systemic disease, a performance status above 2 in almost half of the patients, presence of 'B' symptoms, and increased LDH level in more than two-third of patients. Involvement of extranodal sites was frequent, and a poor IPI was present in more than half of the patients. Interestingly, in patients with long-term HIV suppression, all these adverse prognostic factors were similar to that observed in patients with recently controlled or uncontrolled HIV infection. The spectrum of lymphoma subtypes did not differ with the duration of HIV suppression and was similar to that observed in patients with incomplete suppression of HIV replication from the same cohort or from the literature [14,18,25]. Thirty-six percent of cases were non-IBL-DLBCL and 24% were Burkitt lymphoma. The proportion of Burkitt lymphoma changed negligibly with duration of HIV suppression, despite an increased CD4 cell count at diagnosis from 281 to 357 cells/µl, and was not different from that observed in patients with uncontrolled HIV infection, whereas one could speculate that the proportion of Burkitt lymphoma, which occurs at higher level of CD4 cell count, would be increased in this population [13,17]. Several recent studies [4,6,17] reported stable or even decreased incidence of Burkitt lymphoma in cART era. In contrast, the immunoblastic variant of DLBCL, known to be associated with profound immunodeficiency, has nearly disappeared, representing only 7% of all NHL in this cohort, with only one case after 18 months of HIV suppression. Outside these main subtypes, unusual lymphoproliferative disorders were also present, particularly PEL, which represent 15% of NHL in the study population. Recently described solid PEL-like NHL accounted for five out of the 19 PELs [31,32]. PEL subtype is probably overrepresented in this series, whereas the department is a reference centre for HHV-8-associated lymphoproliferative diseases. However, interestingly, 19 out of the 33 PELs reported in cART era from the HIV lymphoproliferative cohort occurred in patients with HIV suppression.
The results of the present study confirm the possible occurrence of NHL in patients with suppressed HIV infection treated with cART. Pathogenesis of HIV-NHL is not univocal, and several factors are thought to play a role, including not only immune deficiency but also genetic abnormalities, cytokine deregulation, and chronic B-cell stimulation by HIV and/or other viruses in the absence of effective T-cell control [13,34-36].
Most of the malignancies observed in the late cART era are related to infective agents, including hepatitis viruses, human papilloma virus, EBV, and HHV-8, similarly to that observed after solid organ transplantation [37,38]. Several viruses have been associated with lymphomagenesis, including EBV and HHV-8 [13,18,20,21,23,24]. In this study, more than 80% of the tumours were tested for the presence of EBV and HHV-8. More than half of the HIV-NHL arising in this population of patients with undetectable HIV-RNA was still virus associated, without any decline with the duration of HIV suppression. Patients with virus-associated tumour have more frequently a history of Kaposi sarcoma and presented subtypes of lymphoma known to be HHV-8/EBV associated (PEL, LBCL arising in HHV-8 MCD, and IBL-DLCBCL). However, when compared with patients with nonvirus-associated NHL, these patients had no specific characteristics, even though the nadir CD4+ cell count appeared to be slightly lower. In this study, 37% of tumours were associated with EBV, in accordance with literature, in which EBV has been reported to be detected in 25-60% of all HIV-related lymphoma [13,18,23]. This rate did not decrease with longer duration of HIV suppression, and EBV remains present in 43% of NHL arising after long duration of HIV suppression. HHV-8 has been detected in rare lymphoproliferative diseases: PEL, MCD, and LBCL arising in HHV-8 MCD [20,21,24]. In the study population, 23% of NHL was associated with HHV-8, a higher rate than described in the literature, probably reflecting the referral bias mentioned above . However, no significant decrease in HHV-8-associated tumours was observed with the duration of HIV suppression. One limitation of this study is the absence of detection of tumour virus association for the entire NHL cohort of 388 patients, as it was limited to the 128 studied patients, so we could not compare the rate of virus association between patients with and without HIV suppression.
Another limitation of the study is that we could not calculate the incidence rate of NHL in the population of HIV-infected patients with undetectable HIV-RNA. Patients were referred to our department, and we could not evaluate the total population at risk of developing lymphoma. However, in most epidemiological studies, plasma HIV-RNA is not available at the time of NHL diagnosis, and to our knowledge, this incidence has never been evaluated.
HIV-infected patients may still develop NHL after suppression of HIV replication. However, most of these patients had a recent control of HIV infection, and NHL occurred mainly within the 18 months following the suppression of HIV replication. In patients developing NHL after prolonged suppression of HIV replication, no significant impact on the characteristics of NHL was observed, despite a higher CD4 cell count. These NHL remain aggressive, associated with poor prognosis, and the prevalence of virus-associated tumour did not decrease with duration of HIV suppression, despite immune restoration.
Description of the population
Out of the 638 patients with HIV-NHL included in the HIV lymphoproliferative cohort, 388 were enrolled in the cART era (July 1996-September 2008) and 128 (33%) had plasma HIV-RNA below 500 copies/ml at diagnosis of NHL. The patients were followed for a mean period of 32.3 months, with a lost of follow-up rate of 4.7%. The number of cases diagnosed dramatically decreased over time of HIV suppression. For 83 patients (65%), NHL developed within the first 18 months after HIV suppression (Fig. 1).
The baseline characteristics of the 128 study patients are provided in Table 1. The majority of patients were men (87.5%), with a median age of 43 years. At NHL diagnosis, the median CD4 cell count was 297 cells/µl, and the nadir CD4 cell count 112 cells/µl. All but three patients were treated with cART before NHL diagnosis for a median duration of 26.2 months. Regimens included a protease inhibitor in 89 patients, a nonnucleosidic reverse transcriptase inhibitor in 24 patients, and both types of drugs in eight patients. Furthermore, four patients received enfuvirtide or raltegravir. Of the three individuals who developed NHL while not on cART, two received dual ART and one had spontaneous control of HIV replication. The median duration of effective HIV suppression was 10.1 months (IQR 4-27.5 months).
Lymphoma was of the non-IBL-DLBCL type in 46 cases (36%) and Burkitt lymphoma type in 31 cases (24%). More than half of the patients (55%) presented with unfavourable prognostic factor, defined by IPI 3 or 4. Complete remission was achieved in 69% of patients. At follow-up visit, after the end of chemotherapy, the CD4 cell count was 185 cells/µl, and 67% of patients had persistent undetectable plasma HIV-RNA. There were 65 deaths during the follow-up period: 42 (65%) were NHL related, 15 (23%) were treatment related, one was AIDS related, and seven (11%) were from other causes. The median overall survival (OS) was 41.7 months, with a 3-year OS of 52.4% [95% confidence interval (CI) 42.8-61.2].
Analysis was extended to compare these features with those of the 260 patients with NHL and incomplete HIV suppression (plasma HIV-RNA >500 copies/ml) enrolled in the HIV lymphoproliferative cohort during the same period. Patients with suppression of HIV replication were slightly older, had a higher CD4 cell count (297 versus 155 cells/µl, P < 10-4), and were more often treated with cART than patients without suppression of HIV replication. All the NHL characteristics were similar between the two groups.
Comparisons of patients according to the duration of HIV suppression
Characteristics of the patients with long-term HIV suppression, defined as duration of HIV suppression above 18 months were compared with those of patients with recent HIV suppression, defined as a duration of HIV suppression lower than 18 months (Table 1). Date of first undetectable HIV viral load was unknown for three patients, and they could not be included in the following analysis. In the above 18-month group, patients were more frequently women and were older than in the lower than 18-month group. The differences in HIV characteristics were a longer duration of HIV infection, a longer duration of cART, and a higher CD4 cell count at diagnosis of NHL in the above 18-month group. The distribution of the CD4 cell count by year of HIV suppression is presented in Fig. 2, showing that the most striking feature was a rapid decrease in numbers and proportions of patients with a CD4 cell count below 200 cells/µl. Proportion of patients with a CD4 cell count above 350 cells/µl was not different between the lower than 18-month and the above 18-month groups (38.5 versus 50%, P = 0.22). The other HIV characteristics, including the nadir CD4 cell count (P = 0.71), and classes of antiretroviral regimens received at NHL diagnosis (data not shown) were not significantly different between the two groups.
None of the NHL characteristics, including clinical and pathological features, staging, performance status, lacticodehydrogenase (LDH) level, 'B' symptoms, IPI, and therapy regimens used, differed significantly between the two groups (Table 1). A higher CD4 cell count was observed in the above 18-month group for the two main histological groups (377 versus 131 cells/µl for non-IBL-DLBCL and 357 versus 281 cells/µl for Burkitt lymphoma). Response to chemotherapy and relapse rate was not different between the two groups (64% in the over 18-month group versus 72.5% in the less than 18-month group, P = 0.35; and 36% in the over 18-month group versus 21% in the less than 18-month group, P = 0.14). The 3-year OS for the over 18-month group was 46% (95% CI 28.4-61.3%) compared with 56% (95% CI 44.4-66.7%) for the less than 18-month group (log-rank, P = 0.08).
Virus association could be assessed for EBV in 115 tumours (90%) and for HHV-8 in 105 tumours (82%). Fifty-seven patients (52%) presented with a virus-associated tumour. Of these cases, 33 expressed EBV alone, 15 expressed HHV-8 alone, and nine expressed both EBV and HHV-8 (Table 2). Among virus-associated tumours, 22 were DLBCL (39%) and 19 were PEL (33%). Of the 24 HHV-8-associated tumours, 19 were PEL and five large B-cell lymphomas arising in HHV-8-associated multicentric Castleman disease (MCD) (LBCL arising in HHV-8 MCD). All other HIV-related and NHL-related characteristics were not different between the two groups, even though there was a slight trend for a lower nadir CD4 cell count in the virus-associated group (Table 3). The proportion of virus (either EBV or HHV-8)-associated tumour was not associated with the duration of HIV suppression (Table 2). Kaposi sarcoma was more frequent in patients with virus-associated tumour than in patients with nonvirus-associated tumour (22 versus 7.5%, P = 0.02).
The present study included patients with plasma HIV-RNA below 500 copies/ml at the time of NHL diagnosis. All these patients have been enrolled in a single-institution, prospective, observational cohort study (HIV lymphoproliferative cohort) recruiting all consecutive HIV-infected patients with a first episode of lymphoproliferative disorder. All patients with NHL included in this cohort study had confirmed HIV infection and histologically or cytologically proven NHL. The diagnosis of PCNSL was made histologically or clinically on the basis of the computed tomography scan characteristics of the lesion and failure to respond to empirical therapy for cerebral toxoplasmosis.
Collection of data
Standardized data collection forms were completed at NHL diagnosis to provide information about sociodemographic, NHL, and HIV-related characteristics. Lymphomas were pathologically classified according to the WHO classification . Routine staging was performed in all patients, according to the Ann Arbor staging criteria. NHL prognostic score was evaluated by the lymphoma International Prognostic Index (IPI). This index has been found to be prognostic in non-HIV-associated NHL  and in HIV-infected patients . The chemotherapy regimens used were categorized into five groups: intensive (ACVBP, COPADM, and CHOP-Methotrexate), standard (CHOP or CHOP-derived), low dose (mini-CHOP and COP), or absence of treatment. Use of rituximab was also recorded. Individuals were classified as cART users if they had used cART for at least 2 months before NHL diagnosis.
A follow-up visit took place within 2 months after the end of chemotherapy. Complete response was defined as disappearance of all evidence of malignant disease. Partial remission and progressive disease were considered as treatment failures. Change in cART during chemotherapy was recorded, and immunovirological status was evaluated. Thereafter, data concerning vital status and NHL relapse were recorded until death or last available visit for patients lost to follow-up.
Pathological and virological methods
A centralized review of all available pathological specimens of NHL diagnosis was uniformly analysed by a single pathologist (V.M.), and they were classified according to the WHO classification . Solid primary effusion lymphoma (PEL)-like NHL were classified as PEL [31,32]. The presence of EBV was detected by EBV-encoded RNAs (EBERs) in-situ hybridization and latent membrane protein (LMP) immunohistochemistry, and presence of HHV-8 by latency-associated nuclear antigen (LANA) immunohistochemistry. For patients with only cytological samples available, the detection of EBV and HHV-8 was performed using quantitative real-time PCR. Tumour was defined as virus associated if EBV or HHV-8 was detected in the tumour cells.
Suppression of HIV replication was defined as a plasma HIV-RNA level below the mostly used limit of detection since 1996 (500 copies/ml). The duration of HIV suppression was calculated from the first undetectable plasma HIV-RNA sample, without any plasma HIV-RNA measurement above 500 copies/ml, until NHL diagnosis.
Individuals were followed prospectively from enrolment in the cohort until death, last follow-up visit, or 1 September 2008, whichever occurred first. Descriptive analysis used medians with interquartile range (IQR) values. Comparison between groups used Wilcoxon rank-sum test for continuous variables, and Pearson chi-squared or Fisher's exact test for noncontinuous variables. All P values were two-sided. Survival was estimated using the Kaplan-Meier method and was tested using the log-rank test. All analyses were conducted using Stata, version 9.2 (StatCorp Inc., College Station, Texas, USA).