Pembrolizumab Signal in HIV-Negative Kaposi Sarcoma Remains Hypothesis, Not Standard Care
Kaposi sarcoma sits at the intersection of viral oncogenesis, immune dysregulation, and abnormal angiogenesis, which makes immune-checkpoint blockade biologically plausible in HIV-negative disease. The problem is not plausibility. The problem is evidence quality.
Direct clinical data remain limited to a small, non-comparative phase II cohort of pembrolizumab in classic or endemic Kaposi sarcoma. A 71% best objective response rate is an encouraging signal, but it is not comparative evidence and it is not enough to support biomarker-guided treatment selection in routine care.
Why HIV-Negative Kaposi Sarcoma Cannot Be Treated as One Disease
HIV-negative Kaposi sarcoma is an umbrella term that covers biologically different settings. Classic disease usually arises in older men and often follows an indolent cutaneous course. Endemic disease in KSHV-endemic regions can range from slowly progressive skin lesions to aggressive nodal phenotypes. Iatrogenic disease develops under pharmacologic immune suppression, while transplant-associated disease carries the additional problem of graft preservation.
Those differences are not academic. Age, geography, immune reserve, lesion distribution, viral burden, and reversibility of immune suppression all shape both expected benefit and toxicity from checkpoint inhibition. A dataset that mixes these subtypes without stratification can flatten important treatment effects and make a modest signal look either stronger or weaker than it really is.
Limited local disease is still handled locally. Disseminated, symptomatic, rapidly progressive, or visceral disease is where systemic treatment enters the conversation, and that is the clinical space where immunotherapy is being explored.
The Biology Supports PD-1 Blockade, but Biology Is Not Proof
KSHV persists as a latent episome in tumor cells, and latency-associated nuclear antigen helps maintain the viral genome while influencing transcription, proliferation, and survival. Intermittent lytic activity adds cytokine production, endothelial reprogramming, vascular permeability, and angiogenesis to the picture. The lesion that results is not a simple solid tumor; it is a vascular neoplasm built from infected spindle cells, inflammatory infiltrates, and a highly active microenvironment.
Persistent viral infection should, in theory, create non-self antigens that the immune system can recognize. At the same time, KSHV dampens antigen presentation, alters interferon signaling, recruits suppressive immune populations, and promotes T-cell dysfunction. PD-1/PD-L1 signaling is one part of that escape network. Blocking it may restore antitumor responses even when somatic mutation burden is low.
Negative PD-L1 staining does not rule out benefit. Low tumor mutational burden does not rule out benefit either.
The reverse is also true: a lesion with immune exclusion, poor antigen presentation, irreversible T-cell dysfunction, or dominant suppressive myeloid signaling may still resist treatment despite a plausible mechanism. Biology can justify a trial; it cannot substitute for one.
Pembrolizumab Response Does Not Yet Define a Predictive Strategy
The best available clinical signal comes from the small pembrolizumab phase II cohort in classic or endemic Kaposi sarcoma. The 71% best objective response rate is the most concrete efficacy datapoint in this review, but its interpretation has to stay narrow. The cohort was non-comparative, small, and subtype-limited. It cannot establish superiority over existing systemic options, and it cannot validate a biomarker algorithm for selecting patients.
That distinction matters at the bedside. A high response rate in an early cohort can be real and still be misleading if the enrolled patients were easier to treat, more immune competent, or more likely to have cutaneous disease that is easier to measure than visceral or nodal disease. Multifocal plaques and nodules also create response-assessment problems because changes in color, thickness, edema, and inflammation can evolve at different speeds.
A dramatic lesion can soften before a durable antitumor effect has been proven.
PD-L1, Tumor Mutational Burden, and KSHV DNA Are Still Candidate Markers
Several biomarkers have been proposed, but none is ready to stand alone as a selection or exclusion test. PD-L1 expression, KSHV DNA, and tumor mutational burden are the most visible candidates, yet each carries major interpretive problems.
PD-L1 varies by antibody clone, scoring method, threshold, and cellular compartment. Archival tissue may not match the lesion actually treated. Spatial heterogeneity matters because a plaque, nodule, lymph node, and visceral site may not share the same immune biology even when they are all HHV-8 positive. KSHV DNA is also assay-sensitive: specimen type, timing, extraction, and calibration all influence the readout.
A marker that correlates with disease extent is not necessarily predictive of benefit from checkpoint blockade. Prognostic information and predictive information are not interchangeable, especially in a tumor where viral burden, inflammation, and immune escape are tightly entangled.
Why the Review Rejects Stand-Alone Biomarker Use
The review framework is intentionally conservative because the literature is vulnerable to multiple forms of bias. Specialist-center referral favors patients with measurable cutaneous disease, adequate performance status, and capacity for repeated visits or biopsies. Exceptional responses are more likely to be published than primary resistance. Older reports often lack lesion counts, standardized photographs, duration of response, and patient-reported outcomes. Subtype mixing can erase clinically meaningful differences among classic, endemic, iatrogenic, and transplant-associated disease.
Biomarker studies face an additional problem: small cohorts invite overfitting. If a cutoff is chosen after seeing the outcome, it may look impressive in one series and fail immediately in another. That is why the framework emphasizes prespecified assays, locked thresholds, longitudinal sampling, transparent handling of missing data, reproducibility testing, and external validation.
What a Credible Next Study Would Need to Show
Future prospective work has to separate HIV-negative Kaposi sarcoma into meaningful biological and epidemiological subgroups rather than collapsing them into one label. It also needs to test whether a candidate marker identifies benefit from PD-1 blockade, not merely whether the marker is present in more extensive or more inflamed disease.
A valid study would require subtype stratification, standardized response assessment, matched tissue or blood sampling, and explicit accounting for prior therapy and immune suppression. It should also track the lesion type being sampled, since the biology of a cutaneous plaque is not identical to that of a nodal or visceral lesion. Without those controls, biomarker claims will keep outrunning the data.
For now, pembrolizumab remains the clearest immunotherapy signal in HIV-negative Kaposi sarcoma, and PD-L1, KSHV DNA, and tumor mutational burden remain exploratory rather than decision-ready. The evidence base is too small, too heterogeneous, and too methodologically uneven to support routine biomarker-guided treatment selection.