Clinical, virologic, and immunologic outcomes in lymphoma survivors and in cancer-free, HIV-1–infected patients

A matched cohort study

Authors


Abstract

BACKGROUND

The objective of this study was to compare immunologic, virologic, and clinical outcomes between living human immunodeficiency virus (HIV)-infected individuals who had a diagnosis of lymphoma versus outcomes in a control group of cancer-free, HIV-infected patients.

METHODS

In this matched cohort study, patients in the case group were survivors of incident lymphomas that occurred between 1997 and June 2010. Controls were living, cancer-free, HIV-infected patients who were matched to cases at a 4:1 ratio by age, sex, nadir CD4 cell count, and year of HIV diagnosis. The date of lymphoma diagnosis served as the baseline in cases and in the corresponding controls.

RESULTS

In total, 62 patients (cases) who had lymphoma (20 with Hodgkin disease [HD] and 42 with non-Hodgkin lymphoma [NHL]) were compared with 211 controls. The overall median follow-up was 4.8 years (interquartile range, 2.0-7.9 years). The CD4 cell count at baseline was 278 cells/mm3 (interquartile range, 122-419 cells/mm3) in cases versus 421 cells/mm3 (interquartile range, 222-574 cells/mm3) in controls (P = .003). At the last available visit, the CD4 cell count was 412 cells/mm3 (range, 269-694 cells/mm3) in cases versus 518 cells/mm3 (interquartile range, 350-661 cells/mm3) in controls (P = .087). The proportion of patients who achieved virologic success increased from 30% at baseline to 74% at the last available visit in cases (P = .008) and from 51% to 81% in controls (P = .0286). Patients with HD reached higher CD4 cell counts at their last visit than patients with NHL (589 cells/mm3 [range, 400-841 cells/mm3] vs 332 cells/mm3 [interquartile range, 220-530 cells/mm3], respectively; P = .003). Virologic success was similar between patients with HD and patients with NHL at the last visit. Forty cases (65%) and 76 controls (36%) experienced at least 1 clinical event after baseline (P < .0001); cases were associated with a shorter time to occurrence of the first clinical event compared with controls (P < .0001).

CONCLUSIONS

HIV-infected lymphoma survivors experienced more clinical events than controls, especially during the first year of follow-up, but they reached similar long-term immunologic and virologic outcomes. Cancer 2013;119:2710–2719. © 2013 American Cancer Society.

INTRODUCTION

Although highly active antiretroviral therapy (HAART) noticeably reduced mortality and morbidity of human immunodeficiency virus (HIV)-infected patients and lowered the incidence of acquired immunodeficiency syndrome (AIDS)-related cancers,[1-4] lymphomas still remain a major issue in the HIV-infected population.[5, 6] Main concerns regarding the use of chemotherapy (CT) in HIV-infected patients, compared with the general population, are subsequent immune depletion and an increased rate of infections,[7] mostly because, like most non-Hodgkin lymphomas (NHLs), AIDS-related cancers generally occur when the CD4 cell count is low and the patient is severely immunosuppressed.[8] Moreover, different CT regimens used to treat lymphoma have an impact on the development of metabolic comorbidities,[9-11] which may contribute to the premature ageing described in the HIV-infected population.[12, 13]

After the introduction of HAART, immune recovery after CT and autologous stem cell transplantation appeared to approach the recovery observed in immunocompetent patients[14, 15]; however, data on long-term immunologic and virologic outcome as well as the development of clinical complications are lacking. The objective of the current study was to compare, over a 10-year period, HIV viral load suppression, CD4 cell count recovery, and the proportion of clinical events between HIV-infected individuals who survived a lymphoma diagnosis after receiving complete CT treatment and individuals without lymphoma.

MATERIALS AND METHODS

Study Design

This was a matched cohort study of HIV-infected patients recorded in the Infectious Diseases Database of the San Raffaele Hospital (IDD-HSR). This hospital is a multidisciplinary university scientific institute with specialized outpatient facilities and hospital wards; the hospital also includes many research units working on basic science or clinical topics.

The Infectious Diseases Department currently follows more than 4500 HIV-infected patients; within the Department, an observational database (the IDD-HSR; it is included in the most important European cohort networks, such as COHERE, EuroSIDA, D:A:D) collects demographic, clinical, therapeutic, and laboratory data on adult patients who are receiving primary care for HIV infection, as either outpatients or inpatients, at the department. At the first visit to our clinic, patients provide written informed consent to include their clinical and laboratory data in the IDD-HSR for scientific purposes. Clinical data are systematically recorded by infectious diseases specialists working at the clinic using an electronic chart linked to the database.

In the current study, we included cases and controls who were aged >18 years, who were alive, and who were followed in our clinic at time of data cutoff (December 31, 2010). For the case group, we identified HIV-infected patients in the IDD-HSR database who were diagnosed with histologically confirmed, high-grade NHL or Hodgkin disease (HD) since January 1, 1997; who successfully completed CT (within approximately 6 months); who were alive as of December 31, 2010; and who had data available on their date of birth, sex, date of first HIV-positive test, and date of cancer occurrence.

A control sample (n = 211) of living, HIV-infected patients from the same cohort in the IDD-HSR database who did not have any cancer diagnosis (up to December 31, 2010) was matched to cases at a 4:1 ratio by age (±1 year), sex, nadir CD4 count (±50 cells/mm3), year of HIV diagnosis (±1 year), and index year (year of the matched case's lymphoma diagnosis). Forty-three patients with lymphoma were matched with 4 controls, 7 patients were matched with 3 controls, 6 patients were with 2 controls, and 6 patients were matched with 1 control. Age and nadir CD4 count were evaluated at the index date.

An index date (defined as baseline) was assigned to each case and control. For cases, the index date was the date of their lymphoma diagnosis. For controls, the assigned index date was the index date of their corresponding case. In patients who had more than 1 diagnosis of lymphoma, we considered the first event. Follow-up was determined as the time interval between baseline and the last visit for which data were available.

HIV diagnosis was defined as the date of the first HIV-positive test. Virologic success (VS) was defined as the determination of an HIV-1 RNA plasma level <50 copies/mL. Among HIV-infected patients, VS has been possible only after the introduction of HAART; thus, to properly evaluate the proportion of VS in the 2 groups, for the current study, we decided to consider only cases and corresponding controls with an index date after January 1, 1997 (according to the availability of HAART in Italy).

Specific prophylaxis for Pneumocystis jirovecii pneumonia (for a CD4 cell count <200 cells/mm3), Toxoplasma gondii encephalitis (for a CD4 cell count <100 cells/mm3 and positive immunoglobulin G), and Mycobacterium avium complex infection (for a CD4 cell count <50 cells/mm3) was prescribed according to Centers for Disease Control and Prevention (CDC) recommendations (2009[16] and earlier versions[17, 18]) in both cases and controls.

A clinical event was defined as any confirmed diagnosis, observed after baseline and recorded in the IDD-HSR database, of the following categorized diseases: B events and C events according to the 1993 CDC classification,[19] non-HIV–related infectious events, and noninfectious events. For non-HIV–related infectious events, in the current analysis, we considered events that required intravenous therapy and/or hospitalization: sepsis, pneumonia, endocarditis, intra-abdominal infections (including genitourinary tract infections), central nervous system infections, skin and soft tissue infections, and bone infections.

For noninfectious events, we considered metabolic complications grouped as follows: cardiovascular events (myocardial infarction, deep venous thrombosis, pulmonary embolism, and stroke), bone events (osteoporosis [diagnosed according to the Italian guidelines for the diagnosis, prevention, and treatment of osteoporosis[20]] and fractures), first diagnosis of diabetes mellitus type 2 (diagnosed according to American Diabetes Association Criteria[21]), cirrhosis (diagnosed by liver biopsy or elastography), and first diagnosis of chronic kidney disease (defined according to National Kidney Foundation Criteria[22]).

Statistical Analysis

Continuous variables were described by median values and first and third quartiles. Chi-square tests or Mann-Whitney tests were used to compare characteristics of cases and controls.

Variations in CD4 cell counts were calculated as changes from baseline. All CD4 values available during follow-up within each patient were used for the analysis of immunologic changes. Univariable mixed linear models were applied to identify whether some fixed effects (lymphoma occurrence [yes vs no], or type of lymphoma [HD vs NHL], or histologic type of NHL [diffuse large B-cell lymphoma (DLBCL) vs Burkitt lymphoma], or the receipt of rituximab [yes vs no]) that were included alternatively in different models were associated with a mean change in the CD4 cell count from baseline. On the basis of Akaike information criteria, we used an autoregressive order of 1 (AR[1]) covariance structure to control for the correlation within each patient over time. Parameter estimates were obtained by using the maximum likelihood method; mean changes in the CD4 cell count with the corresponding standard error were reported and compared between groups.

For the analysis of VS, HIV-1 RNA values (±30 days) nearest to the considered time points (ie, baseline, 6 months, 1 year, and subsequent 1-year intervals) were used. The Cochran-Armitage test was used to assess the linear trend of VS during follow-up, and the Cochran-Mantel-Haenszel test was calculated to evaluate differences between strata. Differences in the categories of clinical events between cases and controls or between histologic type of lymphoma were evaluated using the chi-square test.

Time to first clinical event distributions were estimated with the Kaplan-Meier method, and comparisons for differences were assessed by using the log-rank test. All statistical tests were 2-sided at the 5% level, and statistical analyses were conducted using the SAS statistical software package (version 9.2; SAS Institute, Inc., Cary, NC).

RESULTS

We compared 62 HIV-infected patients who had lymphoma (20 HD and 42 NHL) with 211 controls. Overall, 273 patients were included in this study and had a median follow-up of 4.8 years (interquartile range, 2.0-7.9 years; cases: 4.6 years [interquartile range, 2.2-8.1 years]; controls: 4.8 years [interquartile range, 2.0-7.9 years]; P = .888). Among cases, the year of lymphoma diagnosis was 2005 (interquartile range, 2001-2007).

Patients received the following CT regimens: 1) all patients with HD received 6 cycles of a doxorubicin, bleomycin, vinblastine, dacarbazine regimen[23, 24]; 2) 33 patients with DLBCL received 6 cycles of cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP),[23, 25] and 11 of those patients also received rituximab according to their CD20-positivite status and CD4 cell count[25]; 3) 9 patients with Burkitt NHL (BURL) received different CT regimens, including 4 who received CHOP (1 with associated rituximab), 3 who received a German Multicenter Acute Lymphoblastic Leukemia (GMALL) protocol,[26] and 2 who received a new, intensive, short-term chemoimmunotherapy regimen.[27]

Two of 33 patients with DLBCL failed on the CHOP regimen and received an additional high-dose CT regimen (rituximab, dexamethasone, cytarabine, and cisplatin [R-DHAP]) and underwent subsequent autologous transplantation.[28] In addition, 1 patient with HD relapsed 4 years after the first episode and also received high-dose R-DHAP and underwent subsequent autologous transplantation.

The characteristics of cases and controls are listed in Table 1. The 2 groups differed in terms of AIDS diagnosis before baseline (P = .029) and years of HAART exposure (P = .046).

Table 1. Case and Control Characteristics
 No. (%) 
CharacteristicCases, n = 62Controls, n = 211P
  1. Abbreviations: ARV, antiretroviral treatment; HAART, highly active antiretroviral treatment; HbsAg, hepatitis B surface antigen; HCVAb, hepatitis C virus antibody; HIV, human immunodeficiency virus; IVDU, intravenous drug user; MSM, men who have sex with men.

  2. a

    This P value was determined with the chi-square test.

  3. b

    This P value was determined with the Mann-Whitney test.

  4. c

    Data were available for 48 cases and 196 controls.

  5. d

    Data were available for 44 cases and 172 controls.

Men57 (92)191 (91).99a
Age: Median [interquartile range], y45.9 [42.3-50.8]45.7 [42.7-50.1].657b
HCVAb-positivec18 (37)75 (38).999a
HbsAg-positived5 (12)12 (7).350a
Diagnosis of AIDS before baseline22 (35)45 (21).029a
HIV risk factor   
IVDU15 (24)58 (27).652a
MSM16 (26)66 (31) 
Heterosexual6 (10)16 (8) 
Other/unknown25 (40)70 (33) 
Years of HIV Median [interquartile range]14.0 [8.7-20.8]15.5 [9.6- 20.8].583b
Years of ARV Median [interquartile range]11.0 [5.2-14.0]12.0 [7.3-14.4].300b
Years of HAART Median [interquartile range]9.2 [5.2-11.8]10.8 [6.4-12.4].046b
No. of available visits per individual: Median [interquartile range]15.5 [8-26]17.0 [8-26].926b
Length of follow-up: Median [interquartile range], y4.6 [2.2-8.1]4.8 [2.0-7.9].888b
CD4 cell count: Median [interquartile range], cells/mm3   
Nadir131 [64-210]142 [50-226].893b
At baseline278 [122-419]421 [222-574].002b
At last visit412 [269-694]518 [350-661].087b
HIV-1 RNA: Median [interquartile range], log10 copies/mL   
At baseline3.26 [1.69-4.77]1.78 [1.69-4.17].009b
At last visit1.69 [1.69-1.73]1.69 [1.69-1.69].285b

At baseline, cases presented with significantly lower CD4 cell counts and a higher viral loads than controls; the proportion of VS was lower in cases than in controls (30% vs 51%, respectively; P = .008). Almost all patients in both groups were receiving antiretroviral treatment at baseline, including 202 controls (96%) and 59 cases (95%; P = .717). Case characteristics according to the type of lymphoma (NHL and HD) are listed in Table 2.

Table 2. Patient Characteristics According to Type of Lymphoma: Hodgkin Disease and Non-Hodgkin Lymphoma
 No. of Patients (%) 
CharacteristicHodgkin Disease, n = 20Non-Hodgkin Disease, n = 42P
  1. Abbreviations: ARV, antiretroviral treatment; HAART, highly active antiretroviral treatment; HbsAg, hepatitis B surface antigen; HCVAb, hepatitis C virus antibody; HIV, human immunodeficiency virus; IVDU, intravenous drug user; MSM, men who have sex with men.

  2. a

    This P value was determined with the Mann-Whitney test.

  3. b

    This P value was determined with the chi-square test.

Age: Median [interquartile range], y46.1 [42.8-50.6]45.9 [41.9-21.1].946a
Men17 (85)40 (95).317b
HCVAb-positive6 (43)12 (35).746b
HbsAg-positive1 (7)4 (14).646b
Diagnosis of AIDS before baseline9 (45)13 (31).395b
HIV risk factor   
IVDU4 (20)11 (26).298b
MSM5 (25)11 (26) 
Heterosexual4 (20)2 (5) 
Other/unknown7 (35)18 (43) 
Years of HIV infection: Median [interquartile range]16.6 [12.3-19.7]13.6 [7.6- 21.1].583a
Years of ARV: Median [interquartile range]13.2 [7.5-15.5]9.5 [4.4-13.5].037a
Years of HAART: Median [interquartile range]10.2 [5.9-12.4]9.2 [4.4-11.5].119a
No. of available visits per patient: Median [interquartile range]13.5 [8.5-27.5]17.0 [7-25].827a
Length of follow-up: Median [interquartile range], y4.4 [2.7-8.7]4.6 [1.7-8.0].668a
CD4 cell count: Median [interquartile range], cells/mm3   
At baseline375 [131-545]245[114-388].258a
At last visit589 [400-841]332 [220-530].003a
HIV-1 RNA: Median [range], log10 copies/mL   
At baseline2.53 [1.69-4.56]3.70 [1.79-4.80].246a
At last visit1.69 [1.69-1.71]1.69 [1.69-1.74].725a

CD4 Recovery

During follow-up, patients with and without lymphoma had different immunologic recoveries. The mean change in CD4 count among cases was 106 ± 24 cells/mm3 compared with 58 ± 13 cells/mm3 among controls (P < .0001). The CD4 count change differed significantly during the first 6 months of follow-up between cases (mean change, −41 ± 12 cells/mm3) compared with controls (mean change, +14 ± 6 cells/mm3; P < .0001). In the first year, the mean CD4 change was −34 ± 15 cells/mm3 in cases and +15 ± 8 cells/mm3 in controls (P = .004). After 1 year, the increase in the CD4 count was 141 ± 34 cells/mm3 among cases and 69 ± 18 cells/mm3 among controls (P = .062), reaching similar long-term CD4 count values (Fig. 1).

Figure 1.

(A) The median change in CD4-positive cell count is illustrated over time. (B) The median CD4-positive cell count is illustrated over time. Below the graphs, the number of available samples (ie, the number of laboratory determinations available between each time interval) and the number of patients (pts) are listed.

Patients with HD and NHL were associated with a different mean change in their CD4 count during follow-up (NHL, 75 ± 34 cells/mm3; HD, 172 ± 48 cells/mm3; P = .011) (Fig. 2). No differences were detected between patients with HD or with NHL compared with their controls regarding mean CD4 changes during follow-up (HD vs corresponding controls, 164 ± 45 cells/mm3 vs 103 ± 28 cells/mm3; P = .251; NHL vs corresponding controls, 70 ± 27 cells/mm3 vs 72 ± 17 cells/mm3; P = .950).

Figure 2.

The median change in CD4-positive cell count over time is illustrated according to the type of lymphoma (Hodgkin disease [HD] or non-Hodgkin lymphoma [NHL]). Below the graph, the number of available samples (ie, the number of laboratory determinations available between each time interval) and the number of patients (pts) are listed.

In patients who developed NHL, we did not observe significant differences between those with DLBCL (n = 33) and those with BURL (n = 9) regarding the baseline CD4 cell count (DLBCL, 217 cells/mm3 [interquartile range, 95-373 cells/mm3]; BURL, 300 cells/mm3 [interquartile range, 260-706 cells/mm3]; P = .06), the mean CD4 change during the first 6 months (DLBCL vs BURL, −50 ± 20 cells/mm3 vs −45 ± 39 cells/mm3; P = .987), or the mean change during follow-up (DLBCL vs BURL, 100 ± 37 cells/mm3 vs −24 ± 70 cells/mm3; P = .267).

In addition, patients with NHL who received rituximab-containing CT regimens (n = 17) presented with higher baseline CD4 cell counts (296 cells/mm3 [interquartile range, 221-550 cells/mm3] vs 186 cells/mm3 [interquartile range, 73-346 cells/mm3]; P = .023) but similar mean CD4 changes during the first 6 months (−40 ± 29 cells/mm3 vs −53 ± 23 cells/mm3; P = .457) and during follow-up (60 ± 60 cells/mm3 vs 96 ± 42 cells/mm3; P = .442) compared with patients who did not receive CT that included rituximab (n = 25).

HIV Viral Suppression

The trend of VS in cases and controls is illustrated in Figure 3. The proportion of patients with VS increased significantly over time in cases (Ptrend < .0001) and controls (Ptrend < .0001; Cochran-Mantel-Haenszel test; P < .0001). The proportion of VS increased from 30% at baseline to 74% at the last available visit in cases and from 51% to 81% in controls (baseline, P = .008; last visit, P = .286).

Figure 3.

The proportion of patients who had virologic success during follow-up is illustrated among cases and controls. Below the graph, the number of available samples (ie, the number of laboratory determinations available between each time interval) and the number of patients (pts) are listed.

When considering the histologic type of lymphoma (HD or NHL), different proportions of VS were observed (Cochran-Mantel-Haenszel test; P < .0001). Among the patients with NHL, the VS proportion increased significantly during follow-up: at baseline, at 1 year, and at the last visit, the VS proportions 24%, 55%, and 74%, respectively (Ptrend < .0001). VS also increased during follow-up among the corresponding controls: the VS proportions at baseline, at 1 year, and at the last visit were 65%, 56%, and 74%, respectively (Ptrend < .0001; Cochran-Mantel-Haenszel test: NHL vs corresponding controls, P < .0001).

The percentage of VS in patients with HD did not increase significantly during follow-up (VS proportion: 40% at baseline, 55% at 1 year, and 74% at the last visit; Ptrend = .735), which differed from what was observed in controls (VS proportion: 50% at baseline, 46% at 1 year, and 85% at the last visit; Ptrend = .0001; Cochran-Mantel-Haenszel test: HD vs corresponding controls, P = .004). No significant difference was detected between patients with HD and patients with NHL regarding VS at the last available visit (P = .920).

In patients who developed NHL, we did not observe significant differences between those with DLBCL (n = 33) and those with BURL (n = 9) regarding the VS proportion at baseline (DLBCL vs BURL: 7 patients [22%] vs 3 patients [33%]; P = .662), at 6 months (DLBCL vs BURL: 21 patients [31%] vs 8 patients [38%]; P = .598), or at the last visit (DLBCL vs BURL: 27 patients [82%] vs 7 patients [78%]; P = .999). In addition, patients with NHL who received a rituximab-containing CT regimen (n = 17) presented with a similar VS proportion at baseline (4 patients [25%] vs 6 patients [24%]; P = .999), at 6 months (15 patients [43%] vs 14 patients [26%]; P = .110), and at the last visit (12 patients [71%] vs 22 patients [88%]; P = .235) compared with patients who did not receive rituximab (n = 25).

Clinical Events

Seventy-four events occurred in cases, and 164 clinical events occurred in controls. The proportion of patients who had at least 1 clinical event differed significantly between cases (40 patients; 65%) and controls (76 patients; 36%; P < .0001). This difference also was statistically significant when we separately considered patients with NHL versus corresponding controls (27 cases [64%] vs 36 controls [38%]; P = .005) and patients with HD versus corresponding controls (13 cases [65%] vs 17 controls [32%]; P = .016); however, no difference was observed between patients with HD and patients with NHL (P = .956). Thirty-nine cases (63%) and 76 controls (36%) were hospitalized for at least 1 event (P = .0002; HD, 13 patients [65%]; NHL, 26 patients [62%]; P = .814).

Similar proportions of B events and C events were observed among cases and controls (at least 1 B event: 2 cases [3.2%] vs 10 controls [4.7%]; P = .999; at least 1 C event: 5 cases [8.1%] vs 19 controls [9%]; P = .999). Higher proportions of cases experienced at least 1 non-HIV–related infectious event than controls (27 cases [44%] vs 45 controls [21%]; P < .0001). No differences in the proportions of patients who had at least 1 noninfectious event were observed between cases and controls (16 cases [25.8%] vs 39 controls [18.5%]; P = .211). No differences were observed between patients with HD and patients with NHL regarding B events (2 patients [10%] vs 0 patients, respectively; P = .10), C events (0 patients [0%] vs 5 patients [12%], respectively; P = .165), non-HIV–related infectious events (9 patients [45%] vs 18 patients [43%], respectively; P = .874), or noninfectious events (7 patients [35%] vs 9 patients [21%], respectively; P = .353).

Cases were associated with a shorter time to first clinical event than controls, as indicated in Figure 4 (P < .0001). Differences between cases and controls were concentrated within the first 6 months: cases experienced non-HIV–related infectious events (17 cases [27%] and 6 controls [3%]; P < .0001) and C events (3 cases [5%] and 1 control [1%]; P = .038) more frequently than controls. No differences in B events or noninfectious events were reported among cases and controls according to the time of occurrence of clinical events.

Figure 4.

The time to occurrence of the first clinical event is illustrated in cases and controls.

Evaluating non-HIV–related infectious events, sepsis was the most frequent event in cases (24 events; 57%), whereas virus infections (labial herpes simplex infections, monometameric herpes zoster, chickenpox, acute viral hepatitis) were the most frequent in controls (21 events; 31%). Among cases, only 1 patient developed a second different malignancy (squamocellular carcinoma of the oral cavity) 2 years after the diagnosis of lymphoma. Finally, we did not observe significant differences in the proportion of clinical events between patients who did or did not receive rituximab (53% vs 72%, respectively; P = .326).

DISCUSSION

In this study, we compared long-term outcomes, including CD4 recovery, VS, and the occurrence of clinical events, between HIV-infected lymphoma survivors and matched HIV-infected patients without lymphoma during a long follow-up period. We observed a significant restoration of the CD4 cell count in HIV-related lymphoma survivors. Although a lower CD4 count was observed in cases at baseline and during the first-year of follow-up, no differences were noted between the 2 groups at subsequent time points. This finding suggests that lymphoma occurrence and subsequent CT did not impair long-term CD4 recovery in our population, as reported by other studies.[14, 15, 29-31] Conversely, we observed a significant, persistent immunologic recovery over a longer follow-up period, confirming that effective antiretroviral treatment leads long-term virologic efficacy, even in patients who receive CT.[32]

With respect to virologic suppression, the overall proportion of VS increased in lymphoma survivors during follow-up and reached proportions comparable to those of controls. Although we observed long-term immunologic recovery and viral suppression, cases experienced more clinical events (P < .0001) than controls. Not surprisingly, cases experienced more infectious events during the first 6 months of follow-up, overlapping the CT administration period. During the same period, cases experienced more non-HIV–related infectious events (such as sepsis) and C events than controls. The higher number of clinical events registered among cases and the shorter time to event occurrence are compatible with a direct role of CT in the occurrence of clinical events among cases because of CT toxicity and its detrimental effect on immune function, as demonstrated previously.[33-35] During follow-up, no other differences in the proportions of clinical events were observed, in line with the immunologic recovery observed.

We might have expected higher proportions of metabolic complications because of CT—in particular, kidney, heart, and liver toxicities36,9,10—and higher incidence of osteoporosis and diabetes because of the use of steroids; however, cases and controls experienced a similar proportion of noninfectious events during the entire follow-up period. Although the closer clinical and radiologic follow-up of patients after a diagnosis of lymphoma than that of controls may have lowered the incidence of preventable clinical events, the lack of any difference observed in the proportion of noninfectious events may have been caused by 1) the very long latency period of this type of pathology, which did not allow us to detect differences between groups, and 2) the relative short-term exposure to possible risk factors for these pathologies (eg steroid use for the occurrence of type 2 diabetes mellitus and osteoporosis). In addition, although noninfectious events have been clearly related to aging, immunodeficiency, and HIV-related factors,[37-39] the impact of age, years of exposure to HIV infection, and nadir CD4 counts on the occurrence of such events in the 2 groups may have been paired, because cases and controls were matched for these factors.

When analyzed separately in patients with HD and NHL, we observed similar CD4 counts at baseline (375 cells/mm3 [interquartile range, 131-545 cells/mm3] vs 245 cells/mm3 [interquartile range, 114-388 cells/mm3], respectively; P = .258); however, patients with HD experienced higher CD4 recovery than patients with NHL at the last available visit (P = .003). This difference may be explained by a more significant T-cell impairment in patients with NHL both because of the different pathogenesis of these lymphomas and because of the different CT regimens. Usually, NHL occurs when the patient is severely immunosuppressed,[40] whereas HD can occur at higher CD4 levels[41]; and some subtypes of NHL, such as Burkitt lymphoma, require high-dose CT.[26, 27, 29] These conditions may explain the lower immune recovery in patients with NHL than in patients with HD.

Among the patients with NHL, immune recovery was not impaired by histologic type (DLBCL or Burkitt lymphoma) or by the receipt of rituximab. Patients who received rituximab presented with higher proportions of clinical events compared with patients who did not receive rituximab, although this difference was not significant, probably because of the small number of patients.

VS was also observed according to the histologic type of lymphoma (HD and NHL). It is noteworthy that, among the patients with NHL, the VS proportion increased significantly during follow-up (Ptrend < .0001), which differed from patients with HD (Ptrend = .735), probably because patients with NHL had a lower proportion of VS at baseline than patients with HD (24% vs 40%). When we compared VS between patients with HD and NHL versus their respective controls, we observed a significant difference in the proportion of VS between patients with HD and their respective controls (P = .004) but no difference in the proportion of VS among patients with NHL and their respective controls. Patients who had HD did not experience significant VS during follow-up, maybe because of their longer antiretroviral exposure, which may limit the availability of efficacious antiretroviral treatment, compared with patients who had NHL.

Some limitations of our study need to be mentioned. First, there is the retrospective nature of this study, which limited our control over the previously collected data; our patients were predominantly men; thus, our results may be generalized only to a predominantly male HIV-infected population. Moreover, only patients who were alive at the time of data cutoff were included as cases; therefore, data on immunovirologic parameters and clinical events were available only for a subgroup of all patients who were diagnosed with lymphoma and received CT, and we excluded all patients who died before the date of data cutoff, independent of the length of their follow-up after a diagnosis of lymphoma. Another limitation is that potential residual confounding factors, other than the diagnosis or treatment of lymphoma, may explain our results, such as the more intensive care of patients with HIV and better patient adherence to antiretroviral therapy. The small number of cases also is a limitation; in particular, the study included only 20 patients with HD. Finally, just few patients reached a follow-up >10 years: the median follow-up was 4.6 years for cases and of 4.8 years for controls.

In conclusion, HIV-infected lymphoma survivors reached similar long-term immunologic recovery and viral suppression compared with HIV-infected patients without lymphoma despite experiencing more clinical events, especially during the first year of follow-up. Although these results confirm the necessity of strict follow-up and patient care, especially during the first year after lymphoma diagnosis, conversely, they provide helpful information to reassure clinicians and patients about the long-term outcome of HIV-infected lymphoma survivors.

FUNDING SUPPORT

No specific funding was disclosed.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary