One report described NHL confined to the CNS in up to 25% of persons with NHL and HIV disease.(5) It is exceedingly rare for these persons, who present with parenchymal brain lesions, to have lymphoma outside the CNS.(13) In contrast to systemic NHL, primary CNS lymphoma in the setting of HIV infection occurs almost exclusively in persons who are severely immunosuppressed.(1,13,14) In all reported series, the median CD4 lymphocyte count in persons with primary CNS lymphoma and HIV disease is less than 50 per microliter.(1,14) The majority of these persons have a history of multiple opportunistic infections.(13-15)

The most common presenting symptoms of primary CNS lymphoma and HIV disease are confusion, lethargy, and memory loss.(16) Other symptoms include hemiparesis, aphasia, seizures, cranial nerve palsies, and headache.(16) Focal neurologic findings do not occur in all patients with primary CNS lymphoma. Single or multiple discrete lesions are the most common findings on computed tomographic (CT) or magnetic resonance imaging (MRI) scans of the brain. The lesions are frequently hypodense and contrast enhancing.

The diagnosis of primary CNS lymphoma is difficult because lymphoma in the brain is difficult to distinguish from intercerebral toxoplasmosis, a common opportunistic infection in patients with HIV disease. CT or MRI scanning offer clues to the diagnosis, but these studies are rarely diagnostic in AIDS patients with CNS lesions. Because MRI scanning reveals a solitary lesion in only 21% of individuals with toxoplasmosis,(15) the finding of a solitary lesion is more suggestive of lymphoma. In individuals with CNS lymphoma, however, MRI demonstrates multiple lesions in 50% of cases,(15) which is less likely to be helpful. Classic radiographic findings of lymphoma, including a periventricular location or a lesion that crosses the midline, occur infrequently in HIV-associated CNS lymphoma.

Thallium-201 SPECT scanning has been used to differentiate these two entities. In one series of 24 patients, all 24 individuals with toxoplasmosis had negative scans, whereas all 12 with CNS lymphoma had positive scans.(17) In another study, 9 individuals with CNS lymphoma had positive findings on thallium SPECT scanning.(18) Of concern, however, was the fact that 3 of 10 patients with toxoplasmosis also had positive scan results. Although data from larger numbers of patients are needed, these observations suggest that SPECT scanning may prove useful in the diagnosis of CNS lymphoma.

Of even greater interest is the detection of EBV DNA sequences in cerebral spinal fluid by using polymerase chain reaction (PCR). De Luca et al. showed that 7 of 8 individuals with documented CNS lymphoma had positive PCR for EBV in CSF.(19) All 11 patients with brain lesions and no lymphoma had a negative result and 21 individuals with AIDS but no CNS lesions also had negative results. In a second series, all 17 individuals with CNS lymphoma were positive for EBV by PCR and 67 of 68 individuals with HIV and no lymphoma were negative.(20) Confirmatory data from larger studies(21) strongly suggest that PCR is a useful means of noninvasive diagnosis of HIV-associated primary CNS lymphoma and may be of particular value when used in conjunction with thallium SPECT scanning and the results of toxoplasma serologic studies.

Brain biopsy is the standard of care for diagnosis of CNS lymphoma. Because treatment outcome hinges on early diagnosis and institution of therapy, toxoplasma-seronegative individuals, who are unlikely to have a diagnosis of toxoplasmosis, should undergo brain biopsy if CSF cytology is negative. Once the diagnosis of CNS lymphoma has been confirmed, a slit lamp examination is performed to rule-out ocular lymphoma before initiation of therapy.

The use of whole brain radiotherapy has been reported. In three series in which therapy resulted in clinical improvement, 62 to 79% of patients had improvement in their neurologic symptoms.(13,22,23) Overall response rates of 52 to 70% have been reported, with complete response rates of 40 to 50%.(1,13,22,23,24,25) Median survival has been 2 to 4.8 months. Most patients die as a result of non-lymphoma complications of their HIV disease.(1,13,22,23,24,25) The proportion of treated patients who die as a result of lymphoma progression (13 to 55%) suggests the need for improved methods of treatment of the lymphoma itself.(1,13,22,23,24,25)

The use of combined modality therapy (chemo- and radiotherapy) has been investigated.(26,27) In one series, 10 individuals had a complete response rate of 88%, and none of these patients died as a result of their CNS lymphoma.(26) The reported median survival of 3.5 months, however, was no better than that reported in studies in which whole brain radiotherapy alone was used. This finding indicates that to improve the management of CNS lymphoma, it is necessary to improve both the management of the lymphoma and the treatment of the underlying HIV disease. Highly effective multiagent antiretroviral therapies may prolong survival of patients with underlying immunodeficiency disease to the extent that better management of the lymphoma may soon become a more critical factor.

Systemic NHL occurs in persons with widely ranging levels of immune function.(1,3,28) Reports describe a median CD4 count for this group between approximately 100 and 180 cells per microliter.(1,3) Seventy-five percent of cases occur in persons with a CD4 lymphocyte count higher than 50 cells per microliter and approximately 30% occur in persons with a CD4 count in excess of 200 cells per microliter.(28) Data from one institution suggest that the level of immune function at the time of diagnosis of lymphoma has declined over time, with the CD4 count falling from a median of more than 200 per microliter in 1987 to approximately 65 per microliter in 1992.(28) This result was consistent with suggestions that NHL might become more prevalent in persons with more severe immunosuppression as persons with HIV disease survive for longer periods because of better treatment for opportunistic infections.(16,29,30) Multicenter AIDS Cohort (MACs) data suggest a gradual increase in the incidence of HIV NHL through 1995.(31)

Extranodal disease is the rule in persons with HIV-associated NHL at a wide variety of sites. In the first large series of HIV-associated lymphomas, Ziegler et al. reported that 95% of patients from several institutions had evidence of extranodal disease, including 42% with CNS involvement and 33% with bone marrow involvement.(5) Similarly, Knowles et al. reported extranodal disease in 87% of 89 persons diagnosed with HIV-associated NHL at a single institution.(8) In this series, the most common sites of disease were the gastrointestinal (GI) tract, CNS, bone marrow, and liver. In a series of 84 HIV-infected patients with NHL at San Francisco General Hospital, Kaplan et al. found that 31% had extranodal sites of disease alone, with no identifiable site of nodal disease.(3) Several reports described stage IV (extensive extranodal involvement) NHL in between 60 and 70% of persons with HIV disease at presentation.(1,3,5,8)

GI NHL is among the most common presentations, occurring in up to 27% of persons with HIV disease and NHL.(32) Although the majority of these cases involve the stomach,(32) reported sites of disease include the oral cavity,(3,32) esophagus, small bowel,(3,9,32) large bowel,(3,32) appendix,(3,32) and anorectum.(3,33) Clinicians must consider the diagnosis, therefore, in persons with evidence of GI bleeding, dysphagia, abdominal pain, rectal pain, or any other chronic GI complaints. In some cases, lymphoma involving the anorectum presents as a perirectal or perianal abscess.(3) Clinicians should always take a biopsy specimen at the time of incision and drainage if there is any suspicion of an underlying neoplasm.

Involvement of the liver and hepatobiliary tree is frequent and patients may present with either intrahepatic or extrahepatic biliary obstruction.(32) In one unusual case, a high-grade NHL occurred within the wall of the common bile duct.(34) Other commonly involved extranodal sites include bone marrow(1,3,5,8,9) and meninges.(1,3,5,8,9)

Several reports have described unusual extranodal sites of NHL in HIV disease, including subcutaneous and soft tissue, epidural space, gingiva, and paranasal sinus.(1,3,5,8,9) Other reports described cardiac and pericardial sites of NHL in persons with HIV in whom respiratory symptoms and arrhythmia were the presenting findings.(35,36) It is clear that NHL should be included in the differential diagnosis in a variety of clinical situations.

In early retrospective studies, factors associated with the underlying immunodeficiency disorder, such as CD4 count, presence or absence of a prior AIDS-defining diagnosis, and performance status, were identified as important predictors of outcome.(1,3) Presence or absence of extranodal disease, particularly bone marrow involvement, was the only tumor-related factor associated with prognosis.

More recent data, such as the prospective data in 192 patients enrolled in the ACTG 142 study,(37) resemble those factors identified in the International Prognostic Index(38) for non-immunodeficiency associated aggressive non-Hodgkin's lymphomas. In ACTG 142, factors associated with poor prognosis included age greater than 35 years, CD4 count of less than 100, history of IV drug use and, for the first time, tumor bulk as measured by stage of disease.(37) Advanced stage III and IV disease were associated with a poor outcome. A retrospective review of 96 patients treated for aggressive HIV-associated lymphomas indicated that bulky tumor, measured by elevation of serum LDH, was associated with a poor clinical outcome.(39)

Initially, dose-intensive regimens were used in an effort to improve outcome, including studies of high-dose methotrexate and high-dose cytosine arabinoside(7) or regimens containing high-dose cyclophosphamide.(3) In general, these treatment regimens were associated with a high risk of death due to opportunistic infection and survival times that were no better than the 5 to 6 months seen with more standard-dose therapy. Patients treated in these studies had median CD4 counts in the mid-100 range.

A large clinical trial of the aggressive NHL-84 regimen that targeted patients with higher CD4 counts (median 227) was associated with a high complete remission rate, a median survival time of 9 months, and a 2-year disease-free survival of 42%.(40) Whether the improved clinical outcome was attributable to the aggressive chemotherapy regimen or to the fact that patients with higher CD4 counts are more likely to survive longer is not clear.

The disappointing results initially observed with higher dose regimens prompted The AIDS Clinical Trials Group to use an mBACOD (methotrexate, bleomycin, doxorubicin, cyclophosphamide, vincristine, dexamethasone) regimen in which the dosages of cyclophosphamide and doxorubicin were reduced by approximately 50% in 35 patients with aggressive AIDS-associated lymphomas.(41) The results -- complete response rate of 46% and median survival of 5.6 months -- were not significantly different from those obtained in trials of standard-dose chemotherapy, but the hematologic toxicity was significantly less. In patients in this study, the median CD4 count was 150. Subsequently, an Italian study of a similar low-dose regimen in 37 patients with lymphoma and a median CD4 count of 25 achieved only a 14% complete remission rate and a median survival of only 3.5 months.(42) In combination, these data imply a more important effect of immune function on survival than choice of therapeutic regimen.

The AIDS Clinical Trials Group subsequently designed a clinical trial (ACTG 142) to directly compare the importance of chemotherapy dose intensity for clinical outcome in AIDS-lymphoma. Patients were randomized to receive either standard-dose mBACOD chemotherapy with GM-CSF support or reduced-dose mBACOD with GM-CSF administered only as required for neutropenia(2) (Table 1). The results (Table 2) showed no significant difference in complete response rate, response duration, time to progression, and overall or disease-free survival. The difference for response duration nearly reached statistical significance in favor of the low-dose regimens.

When ACTG 142 subjects are divided into two groups according to CD4 count (> 100 or < 100 CD4 cells/mm³), those with better immune function survive longer than those with poor immune function. Within each of these two CD4 cohorts, however, no significant differences are observed with respect to treatment assignment, indicating that these results may apply to the majority of individuals with HIV-associated lymphoma and that a dose-reduced chemotherapy regimen is the most appropriate choice for most individuals with AIDS-lymphoma. The number of individuals in this clinical trial with CD4 counts greater than 200/mm3 was insufficient to determine the possible benefit of a standard-dose regimen to those with better immune function. It is therefore recommended that patients with relatively intact immune function (CD4 > 200/mm³) be considered for standard-dose therapy, because they are likely to tolerate full-dose therapy better than those with poor immune function.

In 1996, Sparano et al. studied the effects of continuous infusion of cyclophosphamide, doxorubicin, and etoposide on the management of AIDS-lymphoma.(43) This treatment approach involved a 96-hour continuous infusion of these agents with G-CSF. Patients in group A received adjunctive therapy with didanosine, in cycles 1,2,5, and 6; patients in group B received this antiretroviral agent in cycles 3,4,5, and 6. The complete response rate observed (56%) was not significantly different from that observed in previous trials, although the median survival time of 18.4 months appeared to be significantly longer than that reported previously. It is not clear whether prolonged survival is attributable to improvement in antiviral therapy, more recent improvement in the overall management of HIV disease, or a regimen that truly produces more durable responses. A large phase II study of this chemotherapy regimen is currently underway in the Eastern Cooperative Oncology Group.

Treatment outcome is poor for individuals with lymphoma that is refractory to initial therapy. Tirelli et al. reported on the use of a combination of etoposide, mitoxantrone, and prednimustine in 21 patients with either primary resistance or who relapsed after having complete responses. The complete response rate in this group was 26% and the median survival time was only 2 months. The majority of patients ultimately died of refractory lymphoma.(44)

In a series of 14 patients treated at San Francisco General Hospital with escalating doses of continuous infusion ifosfamide and etoposide, the treatment outcome was similar: the overall complete response rate series was 43%, with a relatively short median response duration of only 79 days.(45)

Levine et al. reported the treatment of 35 patients with refractory AIDS-lymphoma with the single agent mitoguazone (MGBG). Although MGBG appeared to be extremely well tolerated and non-myelosuppressive, the overall response rate of 23% suggests that MGBG is better used in combination regimens than as a single agent.(46)

The largest prospective series reported, the ACTG 142 study of 192 patients, cited an incidence of meningeal lymphoma at diagnosis of only 3% in contrast to a 3 to 25% incidence of lymphomatous meningitis at the time of diagnosis of systemic lymphoma as determined in relatively small retrospective series of patients. The frequency of meningeal relapse is more difficult to determine, principally because most patients in reported series routinely received intrathecal chemotherapeutic prophylaxis. In one small study, however, Gill et al. treated 22 patients with a high-dose methotrexate, high-dose cytosine arabinoside treatment regimen.(7) Although these agents should have adequate CNS penetration, the frequency of meningeal relapse was high (35%). It is noteworthy, however, that 7 of the 8 relapses occurred in individuals who had bone marrow involvement at the time of their initial diagnosis of lymphoma, a known risk factor for meningeal relapse in the non-immunodeficient population. Similarly, in a series by Lowenthal et al., two of three individuals who developed meningeal relapse had bone marrow involvement at the time of diagnosis.(9)

At the San Francisco General Hospital, in place of routine prophylaxis against meningeal occurrence in patients with systemic AIDS lymphoma, we only provide prophylaxis for meningeal lymphoma for individuals with risk factors for meningeal disease identified in the non-immunodeficiency associated lymphoma population: bone marrow involvement, epidural disease, paranasal sinus involvement, and small, non-cleaved histologic pattern.

With the exception of concerns regarding overlapping myelosuppression when AZT is used (AZT use is therefore not recommended), the use of nucleoside analogues while chemotherapy is being administered is of little concern. Few data are available, however, concerning potential drug interactions between the protease inhibitors and agents such as cyclophosphamide and the anthracyclines. Clinical trials are currently investigating these potential drug interactions. Despite this lack of data, it is important that patients continue receiving multi-agent antiretroviral therapies during the administration of chemotherapy. Physicians are urged to watch for evidence of excessive toxicity in their patients when such combinations are employed.

Since the first cases of Hodgkin's disease occurring in the setting of HIV infection were reported early in the HIV epidemic,(54,55) controversy has surrounded the precise relationship of this malignancy to the underlying immunodeficiency state. Initial epidemiologic data suggested that no relationship existed between HIV infection and the occurrence of Hodgkin's disease. Epidemiologic studies performed by Biggar et al. from San Francisco(56) and New York(57) comparing the rates of Hodgkin's disease in these two cities from the pre-AIDS period (1973-1978) compared with those occurring early in the AIDS epidemic revealed no statistically significant increase in the odds ratio for Hodgkin's disease in single young men.

Data from several recently reported epidemiologic studies, however, have consistently demonstrated an increase in the risk for Hodgkin's disease among HIV-seropositive individuals. Hessol et al.(58) determined the incidence of Hodgkin's disease among 6,704 homosexual men enrolled in the San Francisco City Clinic cohort study. Comparisons were made with rates determined in the general population using the SEER cancer registry data. This study found an age-adjusted standardized morbidity ratio 5.0 (95% confidence interval 2.0 to 10.3) for Hodgkin's disease or an excess risk in HIV-infected homosexual men of 19.3 cases per 100,000 person years, suggesting that Hodgkin's disease may be increased in the setting of HIV infection.

Recent data from the Multi-center AIDS Cohort Study demonstrated an approximately fourfold increase in the risk of Hodgkin's disease in HIV-seropositive individuals.(59) The National Cancer Institute's AIDS Cancer Match Registry, which includes a Phase I database of 95,000 AIDS cases encompassing 5 areas of the United States and a Phase II involving 300,00 cases over 15 areas, has demonstrated an approximately 10-fold increase in the risk of Hodgkin's disease in this patient population.(60) All cases underwent pathologic review and were confirmed to represent Hodgkin's disease. Since the relative risk of Hodgkin's disease in the setting of HIV disease is markedly lower than the relative risk for non-Hodgkin's lymphoma, it is not surprising that large populations observed over long periods of follow-up would be required to demonstrate a significant increase in the risk of Hodgkin's disease.

Clinical observations suggest that Hodgkin's disease in a patient with HIV infection has a different natural history and therapeutic outcome when compared to Hodgkin's disease in the general population. Large clinical series consisting of non-immunodeficient individuals with Hodgkin's disease diagnosed in the United States demonstrate nodular sclerosis Hodgkin's disease to be the most commonly encountered histologic subtype, occurring in 52 to 62% of all cases, with mixed cellularity accounting for approximately 24% and lymphocyte depletion in 3 to 6% of all patients.(61) These proportions appear to be significantly shifted in individuals with HIV disease, with nearly all reported series indicating a high prevalence of mixed cellularity disease and a significantly higher frequency of lymphocyte depletion disease.(62-66) In a series of 24 patients reported from a cooperative study of six hospitals,(65) 68% of patients presented with either mixed cellularity or lymphocyte-depleted histology.

Advanced stage Hodgkin's disease (Stage III or IV) also seems to occur with greater frequency in the HIV-infected population compared with the general population. Stage III disease is described in 27% and stage IV in 16 to 26% of non-immunodeficient individuals with Hodgkin's disease.(61) Various series, however, have demonstrated stage III or IV disease in 75 to 94%(66,67) of HIV-seropositive individuals with Hodgkin's disease. In Serrano's series, 91% presented with stage III or IV disease,(65) and in a San Francisco General Hospital series, all but 1 of 14 individuals presented with stage III or IV disease.(68)

Bone marrow appears to be one of the most frequent extranodal sites of involvement in HIV-associated Hodgkin's disease. Serrano et al.(65) identified bone marrow involvement in 10% of 125 HIV-seronegative patients in Spain, compared with 50% of HIV-infected individuals. Although bone marrow may be the only site of disease, a search of other sites of disease should be undertaken as Hodgkin's disease may be extremely difficult to diagnose from bone marrow specimens alone.

The prognosis for patients with HIV-associated Hodgkin's disease is significantly worse than that observed in the general population with Hodgkin's disease. In patients with advanced stage III or IV disease, the ABVD regimen (doxorubicin, bleomycin, vinblastine, decarbazine) has been associated with complete remission in approximately 70 to 80% of patients and with long-term disease-free survival in 60 to 70%.(69) In patients with HIV-associated Hodgkin's disease, complete response rates to a variety of combination chemotherapeutic regimens have ranged from 20 to 100% and median survivals generally in the 11 to 19 month range(62,64,65,66,67,70-72) (Table 3).

Because a variety of combination chemotherapeutic regimens have been employed in the reported series, the optimal treatment regimen remains unclear. In a retrospective evaluation, Tirelli et al. demonstrated an improved clinical outcome in 19 individuals treated with MOPP alternating with or following ABVD compared with 22 who received MOPP alone.(73) A lower rate of opportunistic infections was observed in those receiving MOPP/ABVD.

Errante et al.(71) studied a novel chemotherapy regimen, which included epirubicin, vinblastine, and bleomycin to treat 17 patients with HIV-associated Hodgkin's disease. Patients with a poor performance status received a 50% reduction in the doses of epirubicin and vinblastine, and in both groups, treatment was repeated every 21 days for six cycles. Overall, complete remission was achieved in 53% of the total group; 67% of those with good performance score experienced complete remission, whereas only 20% of those with poor performance experienced complete remission. The median survival of the group as a whole was 11 months. It is unclear whether the poor outcome in those with poor performance status is related to their underlying disease or to the fact that they received inadequate doses of chemotherapy.

Poor bone marrow reserve and the occurrence of opportunistic infections have made it difficult to administer full doses of standard Hodgkin's disease chemotherapy to patients with HIV-associated Hodgkin's disease. In a effort to deliver a relatively non-myelosuppressive combination chemotherapeutic regimen, Kaplan et al.(74) used a combination of bleomycin, vincristine, streptozocin, and etoposide in a small number of patients with HIV-associated Hodgkin's disease. Initial results were encouraging with a high complete response rate and no significant myelosuppression.

More recently, Levine et al. studied the use of standard ABVD chemotherapy combined with GCSF in an effort to reduce the myelosuppression associated with standard chemotherapy.(72) In this trial, 60% of 21 treated individuals achieved complete remission with a median survival of 78 weeks (19 months). Overall, this regimen was tolerated well, although considerable hematologic toxicity did occur.

At this time, available data are not sufficient to allow us to determine which chemotherapeutic regimen may be more efficacious. It is recommended that standard treatment regimens be used for the majority of patients (this will usually be ABVD) with myeloid colony-stimulating factor support. For those individuals with severe immunocompromise (CD < 100/mm³) or poor hematologic reserve, however, consideration may be given to dose-modified standard regimens or for the use of less myelosuppressive chemotherapeutic combinations such as those described above.

HIV InSite is a project of the UCSF Center for HIV Information. Copyright 2009, Regents of the University of California.