A Pilot Study of Liposomal Doxorubicin Combined with Bevacizumab followed by Bevacizumab Monotherapy in Patients with Advanced Kaposi Sarcoma

Kaposi sarcoma is a multifocal angioproliferative tumor caused by Kaposi sarcoma–associated herpes virus (KSHV), also known as human herpesvirus-8 (HHV-8; refs. 1–3). Kaposi sarcoma is a commonly occurring malignancy in sub-Saharan Africa (4), and it is the second most common tumor in people living with HIV in the United States (5). In advanced cases, Kaposi sarcoma may be associated with substantial morbidity, including disfiguring and painful tumors, associated edema, and pleural effusions. In the presence of other KSHV-associated conditions [multicentric Castleman disease (KSHV-MCD) or KSHV-associated inflammatory cytokine syndrome (KICS)], mortality rates for patients with Kaposi sarcoma can be over 50% even with best available therapies (6–8). The goal of Kaposi sarcoma therapy is to induce sustained remission. Antiretroviral therapy (ART) is a cornerstone of treatment for HIV-associated Kaposi sarcoma; it has been reported to yield some response in 20% to 80% of cases, but alone generally is usually inadequate for advanced Kaposi sarcoma (9–11). Several chemotherapeutic agents have been shown to have activity in advanced Kaposi sarcoma, and the most commonly utilized are liposomal doxorubicin and paclitaxel (9, 12–16). Liposomal doxorubicin has been noted to yield response rates ranging from 55% to 70% (13, 15–18). Patients treated with chemotherapy frequently have recurrent disease and often have to be periodically retreated often with the same agents (10); this may be associated with immediate and long-term toxicities, such as cytopenias and potential anthracycline-induced cardiotoxicity.

Abnormal angiogenesis is an early and essential characteristic of Kaposi sarcoma, and involves the production of proangiogenic factors produced by KSHV-infected spindle cells including VEGF-A, basic fibroblast growth factor (bFGF), and platelet-derived growth factor (PDGF; refs. 19–21). Several KSHV gene products, such as viral G protein-coupled receptor (vGPCR), viral IL6, latency-associated nuclear antigen (LANA), and K1 are directly and indirectly involved with upregulation of VEGF-A production, and Kaposi sarcoma tumor samples express increased VEGFR-1 and VEGFR-2 mRNA (19, 22–26). Therefore, targeting angiogenesis in Kaposi sarcoma, and particularly VEGF-A, is a rational therapeutic approach. In addition to its role in tumor angiogenesis, targeting VEGF may have beneficial immune effects. VEGF has been noted to inhibit the maturation of dendritic cells from hematopoietic progenitors, resulting in immunosuppressive effects. Prior studies have shown that antibodies against VEGF, such as bevacizumab, may improve immunologic surveillance and has recently been shown to increase intratumoral CD8⁺ T cells in combination with immunotherapy among patients with renal cell cancer (27, 28).

Bevacizumab is a recombinant humanized monoclonal anti-VEGF-A antibody that is approved and recommended for use in several cancers, including colorectal cancer, non–small cell lung cancer, glioblastoma, and cervical cancer (29, 30). In a phase II study from our group, it was shown to have an acceptable safety profile and activity in patients with advanced Kaposi sarcoma (with or without HIV), with an overall response rate (ORR) of 31% (31). In addition, improvement in tumor-associated edema was observed. Bevacizumab is approved for use with liposomal doxorubicin for recurrent epithelial ovarian, fallopian tube, or primary peritoneal cancer (32). Given the activity of liposomal doxorubicin in advanced Kaposi sarcoma, we sought to determine the safety and response rate of liposomal doxorubicin combined with bevacizumab. Furthermore, we assessed whether bevacizumab alone could maintain a Kaposi sarcoma remission after a limited course of combination therapy with liposomal doxorubicin, thus reducing the cumulative dose of anthracyclines. To this end, we evaluated the response rate and progression-free survival (PFS) for up to six cycles of liposomal doxorubicin combined with bevacizumab, followed by up to 11 cycles of bevacizumab monotherapy in patients with advanced Kaposi sarcoma.

Patients enrolled in this study were adults with pathologically confirmed Kaposi sarcoma and at least five evaluable cutaneous lesions and/or evaluable visceral disease including gastrointestinal or pulmonary Kaposi sarcoma. Adherence to ART was required in patients with HIV and there were no CD4 count exclusion criteria. Patients could have had KSHV-MCD, as long as it was in remission at study entry. Patients required a systolic blood pressure of <150 mm Hg and diastolic blood pressure <90 mm Hg, and those receiving antihypertensive medicines were required to be on a stable regimen for at least 1 month. Additional inclusion criteria included Eastern Cooperative Oncology Group performance status ≤ 2, cardiac ejection fraction >50%, urine protein less than 1+ on dipstick or less than 500 mg on 24-hour collection, absolute neutrophil count >750 cells/μL, platelets >75,000/μL, hemoglobin >9 g/dL. Patients may have received any number of prior therapies, including liposomal doxorubicin or bevacizumab monotherapy. To limit cumulative doxorubicin exposure, patients who had a history of cumulative doxorubicin or liposomal doxorubicin dose of greater than 430 mg/m² were ineligible for the study.

This was an open-label, single-center, pilot study conducted at the NIH Clinical Center. Patients with advanced Kaposi sarcoma (i.e., with one or more indications for systemic therapy) were enrolled in one of two cohorts, with planned enrollment of 10 patients per cohort. Cohort 1 included patients who were HIV-negative (HIV⁻), patients who were HIV positive (HIV⁺) with stable Kaposi sarcoma despite 1 year of ART with viremic control, or patients who were HIV-infected with progressive Kaposi sarcoma despite 4 months of ART. Kaposi sarcoma responses in cohort 1 patients would not be considered as being caused by recent initiation of ART. Cohort 2 included all other patients with advanced HIV–associated Kaposi sarcoma. In cohort 2 patients, it was possible that ART and immune reconstitution contributed to the Kaposi sarcoma response.

In the initial induction phase, patients were treated with a loading dose of bevacizumab 15 mg/kg followed a week later by a combination of bevacizumab 15 mg/kg over 60 minutes and liposomal doxorubicin 20 mg/m². The loading dose was omitted if deferral of chemotherapy by 1 week was deemed unsafe. Patients received the combination of bevacizumab and liposomal doxorubicin every 3 weeks for maximum of six cycles or until a complete response (CR) was obtained. Six cycles were chosen as previous studies have demonstrated that the majority of responses occur within this period and it yields a cumulative dose of liposomal doxorubicin (120 mg/m²), which is substantially below the highest recommended cumulative dose (550 mg/m²; refs. 13, 15–18). Patients with stable disease (SD), partial response (PR), or CR after the induction phase could continue on maintenance bevacizumab every 3 weeks for up to 11 cycles. Patients who had clinical improvement with the combination therapy in the induction phase but subsequently progressed or required further cytotoxic chemotherapy during the maintenance phase could receive a second course of six additional cycles of liposomal doxorubicin with bevacizumab. Patients deriving clinical benefit on bevacizumab after the first or second course of combination therapy could continue therapy for up to a total of 24 months of therapy at investigator discretion. Bevacizumab was provided through a Cooperative Research and Development Agreement between the Center for Cancer Research, NCI, and Genentech.

According to the protocol, bevacizumab therapy was discontinued in the event of life-threatening hemorrhage or thrombotic events. Indications to temporarily hold bevacizumab therapy included systolic blood pressure >160 mm Hg, diastolic blood pressure >100 mm Hg, or proteinuria greater than 2+ on dipstick or 2 g in a 24-hour collection. A cardiac Multiple Uptake Gated Acquisition Scan (MUGA) was done at baseline and repeated after six cycles of therapy or following additional liposomal doxorubicin in combination with bevacizumab. A MUGA was repeated after 3 months if the ejection fraction was <50%. After a grade 4 cerebrovascular ischemic event in one patient, which was attributed to bevacizumab, all patients were commenced on thromboprophylaxis with aspirin 81 mg orally once daily, which was stopped 3 months after the last cycle of therapy.

Patients who were HIV⁺ received ART with adjustments made in accordance with US Department of Health and Human Services guidelines. Pneumocystis jiroveci prophylaxis was commenced in HIV⁺ patients with a CD4 count less than 200 cells/μL, and Mycobacterium avium prophylaxis was considered in those with a CD4 count less than 75 cells/μL. This study was conducted under treatment protocol (ClinicalTrials.gov NCT00923936), and translational study protocol (ClinicalTrials.gov NCT00006518), the latter for a posttreatment biopsy and certain correlative studies. Both protocols were approved by the NCI Institutional Review Board and all patients provided written informed consent in accordance with the Declaration of Helsinki.

Kaposi sarcoma response was evaluated every cycle and categorized as CR, PR, SD or progressive disease (PD) using previously described modified AIDS Clinical Trial Group criteria (14, 33, 34). In patients with HIV-Kaposi sarcoma receiving targeted therapy, improvement may be seen after an initial period of progression; therefore, patients with PD could be treated with bevacizumab at investigator discretion for up to six cycles. Safety was monitored with each cycle of therapy and assessed 4 weeks after the last dose of therapy. Toxicities were graded using NCI Common Terminology Criteria for Adverse Events version 3.0. HIV viral load and CD4 counts were assessed every three cycles.

Correlative assays evaluating select cytokines and growth factors that are either important in Kaposi sarcoma pathogenesis or potentially affected by treatment were performed on stored specimens at baseline, time of best response, and completion of therapy. Serum VEGF-A and cytokines [IL1β, IL6, IL8, IL10, IL12, IFNγ, TNFα, and inducible protein (IP)-10] were evaluated using MSD 96-Well Multiarray Pro-inflammatory 7-plex assay (MesoScale Discovery).

The primary objective was to estimate the ORR, composed of CR plus PR, during up to six cycles of the combination of liposomal doxorubicin and bevacizumab in patients with advanced Kaposi sarcoma in two cohorts of patients. Patients were considered evaluable for response if they received at least two cycles of liposomal doxorubicin or came off the protocol for reasons of treatment toxicity. Secondary objectives included an assessment of the safety and toxicity of the combination therapy in patients with advanced Kaposi sarcoma, assessment of the impact of the combination therapy on CD4 counts, assessment of the rate of CR and the median number of cycles required to achieve PR, and an estimate of the PFS.

Exploratory analyses of the differences in the serum VEGF and cytokines between baseline and best response or between baseline and six cycles of treatment were assessed by a Wilcoxon signed-rank test. Given the number of comparisons made, to be conservative, a P < 0.005 would be considered statistically significant, while 0.005 < P < 0.05 would be considered evidence of a strong trend. The effects of therapy on CD4 counts were evaluated separately in each cohort, as patients in cohort 2 were expected to have increases in their CD4 count when started on an optimized ART regimen. PFS for all patients was determined using the Kaplan–Meier (KM) methods from treatment initiation at baseline to progression. Progression was assessed from best response measurements that were sustained for two successive measurements or, for patients that did not achieve a PR, from baseline measurements; PFS was defined at the time from initiation of therapy to progression as defined above. As liposomal doxorubicin could be added back for patients who did not progress but were felt to benefit from further combination therapy, patients were censored for PFS measurements when liposomal doxorubicin was added back to bevacizumab if this was done before formal progression was documented. Duration of response was calculated among responders and defined as time from response on treatment until progression above baseline measurements over all treatments (first and second course of combination treatments and maintenance therapy) and follow-up.

Between April 2009 and May 2015, 16 male patients with Kaposi sarcoma were enrolled (Table 1), 10 in cohort 1, and six in cohort 2. Enrollment was stopped following completion of cohort 1. Eight patients were black and the median age was 46 years (range, 25–58 years). Of the 10 patients in cohort 1, 2 were HIV⁻ and the remaining eight patients were HIV⁺ and had suppressed HIV below the level of detection at entry. Cohort 1 patients on ART had received it for a median of 45 months (range 9.2–137.4 months) prior to entry. Cohort 2 consisted of five patients with T1 Kaposi sarcoma and one patient with progressive T0 Kaposi sarcoma; this latter patient had an associated low CD4 count and history of opportunistic infections and had restarted ART less than 4 months prior to entering the study. Four of these six patients had detectable HIV RNA at baseline (range, 55–58,660 copies/mL) and two patients had undetectable (<50 copies/mL) HIV VL; patients in cohort 2 had been on ART for a median duration of 2.7 months. The median CD4 count for HIV⁺ patients was 358 cells/μL (range, 79–944 cells/μL). A CD4 count of less than 200 cells/μL was noted in 30% of patients in cohort 1 and 33% of patients in cohort 2 at the time of study enrollment. Eight patients had poor prognosis Kaposi sarcoma (T1S1; ref. 35): six (60%) in cohort 1 and two (33%) in cohort 2. Eight patients in cohort 1 were previously treated for Kaposi sarcoma with cytotoxic chemotherapies, including liposomal doxorubicin (n = 7) and paclitaxel (n = 2). The median prior cumulative dose of liposomal doxorubicin was 120 mg/m² (range, 80–400 mg/m²). Other therapies included bevacizumab alone (n = 1) and immunomodulatory drugs [IFNα (n = 3), pomalidomide (n = 2), or lenalidomide (n = 1)]. Two patients in cohort 1 had received treatment with rituximab for a previous diagnosis of multicentric Castleman disease. The median time from last therapy to study enrollment overall was 9.8 months (range, 0.9–94.3) in those who had therapy besides ART. In cohort 1 20% of patients (n = 2) and in cohort 2 83% of patients (n = 5) had received ART therapy alone.

A total 210 cycles of treatment were administered among all participants (124 cycles of combination liposomal doxorubicin and bevacizumab, plus 86 cycles of bevacizumab monotherapy). Nine patients (cohort 1: five patients, cohort 2: four patients) were retreated with a further six cycles of bevacizumab and liposomal doxorubicin due to PD or Kaposi sarcoma that was felt during maintenance bevacizumab to require additional cytotoxic chemotherapy (Supplementary Fig. S1). Two patients did not receive the loading dose of bevacizumab due to concerns about progressive pulmonary Kaposi sarcoma. Assessable patients received a median of nine cycles of liposomal doxorubicin in the combination phase (range, 1–12 cycles) dosed at 20 mg/m² and 14 cycles of bevacizumab (in combination or maintenance therapy, range, 1–25 cycles). Participants were observed for a median of 16 months (range, 1–77 months). All HIV⁺ patients remained on ART therapy during treatment and had an undetectable viral load at the end of study participation.

After the first six cycles of liposomal doxorubicin, the best response observed in either cohort was a PR (Table 2; Fig. 1). The primary objective of the study was the ORR of the first six cycles of combination therapy, including all patients as evaluable who either completed two cycles of therapy or came off study during that period for toxicity. One patient came off during the first two cycles for toxicity but was still included in the ORR. Using this definition, the ORR was 50% [80% confidence interval (CI), 27%–73%] in cohort 1 and 67% (80% CI, 33%–91%) in cohort 2. Combining both cohorts, the ORR for the 16 patients was 56% (80% CI, 38%–74%). The median time to PR was two cycles (range, 1–6 cycles) in cohort 1 and four cycles (range, 1–5 cycles) in cohort 2. In 11 patients with edema at baseline, 90% had a reduction in tumor-associated edema during the first six cycles of therapy, with reduction in calf measurements ranging from 0.5 cm to 7 cm.

Thirteen patients went on to receive maintenance bevacizumab. One went on to have a CR, and two maintained their PR but came off due to new symptoms (a benign necrotic chest lesion in one, and hyperesthesia in the other, possibly related to bevacizumab). Of the other 10 patients, five had progression from best response during (or in one case at the start of) maintenance, and five patients were given a second course of combination therapy for continued nodular lesions, worsening edema, or symptomatic pulmonary Kaposi sarcoma even though they did not meet criteria for progression. One patient with asymptomatic progressive Kaposi sarcoma did not receive the second course of combination therapy due to a nationwide shortage of liposomal doxorubicin. Median PFS (progression from best response) as assessed during this first course of combination therapy followed by maintenance therapy was 6.9 months (95% CI, 4.5 months—not estimable; Fig. 2).

Nine patients went on to receive a second course of combination therapy because they progressed on maintenance therapy (4) or had sufficiently severe Kaposi sarcoma that they were felt to benefit from additional combination therapy (5). During the second course of combination therapy, three of the nine patients had a new PR as assessed from the initiation of the second course, four had SD, and two had PD (Table 2). Six of these nine patients then went onto a second course of maintenance bevacizumab for a median of five cycles (range, 1–11), and three of these patients achieved a CR following the second course of maintenance. The four patients who had a CR, including the one with a CR after the first course of maintenance, were all from cohort 2. These four patients had a PR after the first course of therapy and had no further recurrence of Kaposi sarcoma after a median follow-up of 48 months. Among the nine patients who had a PR to the initial combination regimen, the median duration of response was not reached and the probability of maintaining a response was 89% at 2 years and 78% at 3 years (Supplementary Fig. S2).

Treatment of three patients with liposomal doxorubicin in combination with bevacizumab had to be discontinued as a result of adverse events within the first six cycles of combination therapy (Table 3). One patient had a grade 4 cerebrovascular ischemia following the first cycle of treatment. This manifested itself as expressive dysphasia and visual disturbance. Imaging demonstrated a large attenuated wedge-shaped infarct and smaller lesions suggestive of acute ischemic infarcts that were felt to be likely related to bevacizumab. The second patient with KS limited to the skin developed rectal bleeding after two cycles of combination therapy. This patient presented with an acute drop in his hemoglobin, and colonoscopy demonstrated evidence of rectal ulcers, which were identified as the cause of bleeding. A third patient discontinued treatment after four cycles due to moderate symptoms of periodontitis and a palatal abscess requiring dental intervention. Milder periodontal disease and/or other dental diseases were noted in six additional patients; two patients discontinued treatment due to periodontal disease during the second course of combination treatment. Eleven of the 16 patients (69%) developed grades 2 to 4 neutropenia, which was more than was seen on the previous phase II study of bevacizumab alone; this increase is likely at least in part attributable to the liposomal doxorubicin (31). Other common adverse events included hypertension and proteinuria (Table 3). Five patients had grade 3 hypertension and were administered one or more antihypertensive medications. One patient required all three medications during the study. After cessation of bevacizumab, all patients remained on one antihypertensive medication and were normotensive on follow-up study assessments. The median ejection fraction among all patients was 61% and there were no significant changes in cardiac function during the study.

The CD4 count among cohort 1 and 2 patients were similar at baseline. There was no significant change in the CD4 count levels between baseline to six cycles of combination treatment overall and in either cohort (Supplementary Table S1). Also, there was no statistically significant difference in the changes in CD4 counts between responders and nonresponders in the study. Overall, HIV VL did not change over the course of treatment and all patients had an undetectable viral load at the end of treatment.

The change in baseline to the end of therapy and to best response of serum VEGF-A and cytokines associated with Kaposi sarcoma pathogenesis were assessed in this study (Table 4). There was a statistically significant decrease in the VEGF-A levels from baseline and after six cycles of combination therapy (median paired difference 92 pg/mL; P = 0.0007) and from baseline to best response within the first six cycles (median paired difference 88.3 pg/mL; P = 0.0001). There were no significant differences in changes in the other cytokines that were analyzed in the study. There was also no difference in the change in VEGF-A levels among patients who received six cycles or less of liposomal doxorubicin as compared with patients who received more than six cycles of chemotherapy [median paired difference 131 pg/mL (six cycles or less) vs. 68 pg/mL (more than six cycles); P = 0.23]. We assessed the difference in cytokine levels from baseline to six cycles of combination treatment between responders and nonresponders and noted that changes VEGF-A levels were not statistically significant between these groups (P = 0.35; Supplementary Table S2). However, decreases in IP-10 and IFNγ levels were greater among responders as compared with nonresponders with trends toward statistical significance (IP-10, P = 0.03 and IFNγ, P = 0.05).

This pilot study evaluated the toxicity and potential benefits of liposomal doxorubicin in combination with bevacizumab followed by bevacizumab monotherapy in patients with classic Kaposi sarcoma, HIV-associated Kaposi sarcoma that did not respond to ART, or patients with advanced HIV-Kaposi sarcoma. Overall, nine of 16 patients entered (56%) had a partial response during the initial course of combination therapy. The majority of patients had six cycles of combination therapy and then started on maintenance bevacizumab. However, nine of the 13 patients who went to maintenance therapy had evidence during maintenance of progression or Kaposi sarcoma that was assessed as sufficiently severe to warrant additional combination therapy and were given a second course of bevacizumab and liposomal doxorubicin. Four patients (all in cohort 2) had a complete response and have not required any further Kaposi sarcoma therapy. Finally, three of the 16 patients had serious adverse events attributable to bevacizumab and resulting in cessation of therapy.

In our previous study of bevacizumab monotherapy, we observed an overall response rate of 31% and the median time to best response in responders was 5 months (31). The aim of this study was to investigate whether the combination of liposomal doxorubicin and bevacizumab would lead to a higher ORR, rapid responses, and that the continued bevacizumab monotherapy might then allow continued control of Kaposi sarcoma while reducing the amount of cumulative anthracycline exposure. Comparison of these two studies is difficult due to factors such as the extent of Kaposi sarcoma, the immune suppression status of patients and prior Kaposi sarcoma therapy. Overall, the proportion of patients with advanced Kaposi sarcoma in on this study was high as compared with the bevacizumab monotherapy study conducted by our group (T1 stage; 88% vs. 76%). The response rate from this study was somewhat higher (56% vs. 31%) with a shorter time to response (1.5 months vs. 5 months) and four patients (25%) attained a CR on this study. This suggests that bevacizumab in combination with liposomal doxorubicin may improve Kaposi sarcoma responses as compared with bevacizumab alone. It is worth noting that unlike our study of bevacizumab monotherapy, in which CD4 counts increased in most patients, most participants in this study had no change in CD4 counts, likely due to administration of liposomal doxorubicin. This suggests that immune restoration may not have contributed to the responses in this study, and highlights the problem of cumulative immune toxicities with chemotherapy used to treat Kaposi sarcoma.

A related question is whether the combination is better than liposomal doxorubicin alone; however, it is challenging to compare the results of this study with trials assessing efficacy of liposomal doxorubicin due to the different patient populations studied, use of ART, and variable responses reported in trials of liposomal doxorubicin (15–17, 36). As compared with these prior studies, patients in this study, especially those in cohort 1, had more prior chemotherapy, including liposomal doxorubicin, implying more resistant, poor risk disease, and higher baseline CD4 counts, indicating that the role of immune reconstitution in the resulting responses may be less significant.

The study also investigated the role of maintenance bevacizumab following six cycles of combination therapy. Responses among 13 patients who commenced bevacizumab maintenance monotherapy were mixed: one had persistent PD from baseline response, three required no further treatment and maintained PR or attained a CR, whereas nine patients had either progression from best response on therapy or had changes in Kaposi sarcoma that warranted further combination therapy. The median PFS within the first course of treatment (combination followed by maintenance) was 6.9 months suggesting reasonable durability of response with this regimen. Although bevacizumab monotherapy had some activity, this study demonstrated that it may not be effective at inducing long-term continued or new Kaposi sarcoma responses in the majority of patients after initial combination therapy. At the same time, it may have enabled longer control of Kaposi sarcoma over time with a reduced cumulative dose of anthracyclines and its associated toxicities, such as cardiotoxicity, which was not seen among patients treated on this study.

Given that Kaposi sarcoma is characterized by angiogenesis, it is worth considering why bevacizumab monotherapy did not yield a higher response rate or sustained improvement. Several factors may contribute, including the existence of redundant angiogenic pathways involved in Kaposi sarcoma pathogenesis and expression on spindle cells of other receptors for angiogenic factors, such as VEGF-C and PDGF (21, 24). However, Kaposi sarcoma–related edema appeared to respond well to bevacizumab treatment, and this may be related to the effect of VEGF in causing fluid leakage. This effect on reduction of edema, seen in 90% of patients with baseline edema, may be associated with improved quality of life in patients with Kaposi sarcoma, even though reduced edema is not in itself a standard criterion for Kaposi sarcoma responses (34). In our previous study, treatment with bevacizumab alone resulted in improvement of edema in 72% of patients. Overall, the probability of maintaining a response among the nine patients with a PR after the first course was 89% at 2 years and 78% at 3 years. In addition, four out of six patients in cohort 2 attained a complete response that did not require further treatment, suggesting that this regimen can yield durable benefits in a substantial subset of patients who initially respond. It also suggests that bevacizumab maintenance may be most effective in patients who have had a reduction in their Kaposi sarcoma tumor burden.

A number of changes in the levels of VEGF-A and other inflammatory cytokines were observed during the course of the study. There was a statistically significant decrease in VEGF-A levels in both cohorts from baseline to the time of best response and the end of treatment (31). These results must be viewed with the understanding that binding of bevacizumab to VEGF-A may affect its measurement. However, changes in VEGF-A were not noted in our bevacizumab monotherapy study despite using the same assay and similar timepoints (31). This observation also varies from other studies that have shown an increase in plasma VEGF-A levels following administration of bevacizumab (37). Interestingly, decreases in serum VEGF levels have been reported following treatment with anthracycline-based regimens without bevacizumab in patients with breast or non-Hodgkin lymphoma (38, 39). This may suggest that the difference is driven by chemotherapy rather than the bevacizumab. Overall, use of circulating VEGF-A as a biomarker in clinical studies have been unpredictable and its use as a biomarker in Kaposi sarcoma requires further validation.

Bevacizumab monotherapy was recently incorporated into treatment guidelines for HIV-related Kaposi sarcoma (40) In the prior monotherapy study, the median time to CR or PR was 5 months, which was longer than responses seen in this study with combination therapy (6–12 weeks patients in either cohort). However, bevacizumab is not without toxicity, and as seen in the current study, use of this regimen must be weighed against the risk of serious adverse events as seen in some patients in this study. The cases of rectal bleeding and cerebrovascular event led to significant morbidity in two patients and were not observed in our prior study of bevacizumab. Other toxicities noted in this study were similar to adverse events in trials of combination bevacizumab and chemotherapy for other malignancies, such as ovarian cancer, where gastrointestinal perforation was seen in 2% of patients (32). After the initial excitement surrounding targeted angiogenic inhibitors, toxicities and fatal adverse events associated with this class of therapies have raised some concern in the oncology community (41, 42). This study highlights the concept that in patients with HIV, the toxicities are most likely associated to the class of therapy rather than issues related to HIV and its impact on the immune system.

Further research would be needed to determine the potential role of bevacizumab and liposomal doxorubicin in the treatment of Kaposi sarcoma. Probably, the most effective approach would be a randomized study comparing bevacizumab in combination with liposomal doxorubicin to liposomal doxorubicin alone. Bevacizumab maintenance could be considered in patients who achieve a response. Such a trial would ideally be done in a more homogeneous population of patients than this study, for example, those with HIV-associated Kaposi sarcoma with well-controlled HIV and prior Kaposi sarcoma treatment. Such a study could specifically stratify treatment by the grade of peripheral edema attributed to Kaposi sarcoma and incorporate edema in response assessments.

In summary, combination therapy with bevacizumab and liposomal doxorubicin followed by maintenance bevacizumab has activity in patients with Kaposi sarcoma, particularly in patients with significant chronic edema associated with Kaposi sarcoma. However, these observations require cautious interpretation in practice; potential toxicities need to be carefully considered, and randomized comparison studies would be needed to assess the utility of this approach as compared with other existing and promising regimens.

R. Ramaswami, M.N. Polizzotto, T.S. Uldrick, K. Lurain, and R. Yarchoan report receiving research support from Celgene through a CRADA at the NCI. R. Ramaswami, T.S. Uldrick, K. Lurain, and R. Yarchoan report receiving drug for a clinical trial from Merck through a CRADA with the NCI. T.S. Uldrick reports receiving other commercial research support from Roche through a CTA with Fred Hutchinson Cancer Research Center. M.N. Polizzotto reports receiving commercial research grants to his institution from Celgene SARL, ViiV Pharmaceuticals, and Janssen Pharmaceuticals, and reports receiving speakers bureau honoraria from Gilead Pharmaceuticals and Celgene SARL. T.S. Uldrick, M.N. Polizzotto, R. Yarchoan, and D. Whitby are co-inventors on US Patent 10,001,483 entitled “Methods for the treatment of Kaposi's sarcoma or KSHV-induced lymphoma using immunomodulatory compounds, and uses of biomarkers.” R. Yarchoan is also a coinventor on patents on a peptide vaccine for HIV and on the treatment of Kaposi sarcoma with IL12, and an immediate family member of R. Yarchoan is a co-inventor on patents related to internalization of target receptors, on KSHV viral IL6, and on the use of calreticulin and calreticulin fragments to inhibit angiogenesis. All rights, title, and interest to these patents have been or should by law be assigned to the U.S. Department of Health and Human Services; the government conveys a portion of the royalties it receives to its employee inventors under the Federal Technology Transfer Act of 1986 (P.L. 99-502). No potential conflicts of interest were disclosed by the other authors.

Conception and design: T.S. Uldrick, S.M. Steinberg, G. Tosato, D. Whitby, R. Yarchoan

Development of methodology: T.S. Uldrick, W.D. Figg

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Ramaswami, T.S. Uldrick, M.N. Polizzotto, K.M. Wyvill, P. Goncalves, A. Widell, K. Lurain, W.D. Figg, R. Yarchoan

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Ramaswami, T.S. Uldrick, M.N. Polizzotto, K. Lurain, S.M. Steinberg, W.D. Figg, G. Tosato, R. Yarchoan

Writing, review, and/or revision of the manuscript: R. Ramaswami, T.S. Uldrick, M.N. Polizzotto, P. Goncalves, K. Lurain, S.M. Steinberg, W.D. Figg, G. Tosato, D. Whitby, R. Yarchoan

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Ramaswami, T.S. Uldrick, R. Yarchoan

Study supervision: T.S. Uldrick, R. Yarchoan

This research was supported by the Intramural Research Program of the NIH, National Cancer Institute and in part by the Intramural Program of the National Cancer Institute, NIH, Department of Health and Human Services (contract HHSN261200800001E). The authors thank the individuals who volunteered and the medical, nursing, pharmacy, data management, and support staffs of the National Cancer Institute and the NIH Clinical Center. We also thank Ms. Kirsta Waldon for her work in coordinating the trial and Dr. Randy Stevens and Dr. Adam Rupert in Leidos, Frederick, MD for performing the cytokine assays.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.