進行性多巣性白質脳症

Bias in clinical decision-making is commonly understood as arising from negative stereotypes or cognitive shortcuts that undermine patient-centered care. Less attention has been paid to forms of bias that emerge from clinicians' well-intentioned efforts to alleviate suffering. This article introduces the concept of compassion bias, defined as a distortion in clinical judgment that occurs when empathy and a desire to avoid harm unintentionally constrain objectivity, patient autonomy, or the range of treatment options offered. We present the case of a young woman with advanced neurologic decline secondary to HIV-associated progressive multifocal leukoencephalopathy, in whom artificial enteral nutrition was initially deferred based on clinicians' assumptions regarding quality of life and suffering. Despite good intentions, this approach delayed engagement with the patient's legal surrogate and limited exploration of family values regarding life-prolonging care, as well as likely shortening her life. Through a structured ethical analysis, we examine how compassion bias interacted with cognitive heuristics, implicit bias, and moral distress, contributing to paternalistic decision-making. Drawing on literature from cognitive bias, disability ethics, and quality-of-life assessment, we argue that compassion bias represents a distinct and underrecognized ethical challenge in hospital-based care. We propose that recognizing compassion bias requires deliberate engagement in reflective, Type 2 reasoning, attention to surrogate decision-making processes, and early incorporation of palliative care to support both families and clinicians. By naming and critically examining compassion bias, clinicians and ethics consultants may better balance beneficence and nonmaleficence while preserving autonomy, trust, and ethically sound decision-making in serious illness care.

PURPOSE OF REVIEW: With fifteen agents now FDA-approved and indications expanding into earlier treatment lines, antibody-drug conjugates (ADCs) represent a rapidly growing class of targeted cancer therapeutics. Despite their selective mechanism of action, infectious complications are clinically significant and have been flagged as a disproportionate safety signal in post-marketing surveillance. This review characterizes the infection risk profiles of all fifteen FDA-approved ADCs through systematic extraction of prescribing information and pivotal trial safety data, and examines the pathophysiological mechanisms, antigen-specific clinical patterns, and evidence-based prophylaxis strategies applicable to this class. RECENT FINDINGS: Calicheamicin-based agents carry the highest infection burden, with grade ≥ 3 infection rates exceeding 30-47% and near-universal severe neutropenia. CD30- and CD79b-targeting vedotin conjugates are associated with opportunistic infections-including progressive multifocal leukoencephalopathy and Pneumocystis jirovecii pneumonia-through mechanisms beyond myelosuppression, particularly T-cell immune surveillance disruption and bystander-effect lymphotoxicity. Solid tumor ADCs demonstrate lower overall infection rates with distinct organ-specific patterns: genitourinary infections predominate with Nectin-4- and Tissue Factor-directed agents, whereas pulmonary events characterize HER2- and c-Met-targeted conjugates. Linker cleavability, drug-to-antibody ratio, and payload metabolism are identified as key pharmacological determinants of myelosuppressive and infectious risk. Infectious complications of ADC therapy are clinically significant but heterogeneous, with risk profiles primarily determined by target antigen, immunologic context, and concurrent treatment. These findings support the growing adoption of a combined prevention strategy incorporating disease-based risk stratification and drug-directed infection prophylaxis.

The human JC polyomavirus (JCPyV) causes the fatal demyelinating disease progressive multifocal leukoencephalopathy (PML) in immunocompromised individuals. JCPyV frequently undergoes mutations in PML patients, and these variants are thought to establish "antibody recognition holes" that enable the virus to circumvent the antibody response of infected individuals. Many of these PML-associated mutations cluster in the glycan receptor-binding site of the virus. Using X-ray crystallography, we investigated the binding modes of JCPyV VP1-specific human monoclonal antibodies (mAbs) that were isolated from healthy donors and individuals who recovered from PML, and that can recognize a panel of PML-associated JCPyV variants. Our structural analyses show that three out of four of these mAbs bind epitopes that overlap with the glycan-receptor binding site at the surface of the virus particle. The observed interactions explain how PML-associated mutations strategically interfere with antibody recognition, resulting in immune evasion. In contrast, mAb 29B1 engages a region of the capsid that is distant from the glycan site and does not feature mutations associated with PML. The binding site is conserved in the closely related BK polyomavirus (BKPyV), and we show that mAb 29B1 binds both viruses with high affinity and blocks infection. Our findings form an excellent platform for the development of therapeutic Ab approaches and potential vaccination strategies that could protect at-risk patients from infections with both JCPyV and BKPyV. Moreover, small molecules that target the mAb 29B1 binding site could be potentially effective against both viruses.

BACKGROUND: Progressive multifocal leukoencephalopathy (PML) is a fatal demyelinating disease caused by JC polyomavirus (JCV) that affects immunosuppressed individuals and has no approved therapy. Here we evaluate the safety and efficacy of third-party, off-the-shelf, partially HLA-matched BK virus-specific T cells (BK-VST) in patients with PML. METHODS: We conducted a single-center phase 2 study where 37 patients received most closely HLA-matched BK-VST at 2.0 x 105 T cells/kg, with additional infusions permitted for subjects who did not achieve a complete response. Twenty-three patients had an underlying diagnosis of hematologic malignancy. Each patient received a median of 2 doses (range, 1-12; IQR 2). Overall response (OR) was defined as virologic clearance with neurologic stabilization or improvement. RESULTS: During 12 months of follow-up, 21 patients (56.8%) achieved an OR, including 16 (43.2%) with a complete response. Responses were rapid (median 23 days) and durable, with no loss of response observed despite limited HLA matching. The 1-year overall survival was 61.1% (95% CI, 41.9-77.4). Patients achieving an OR after the first infusion had markedly improved 1-year survival compared with non-responders (93.3% vs 0%; p<0.001). Higher donor IL-2 and IFN-γ-producing T-cell frequencies and greater HLA matching correlated with clinical benefit. CONCLUSIONS: These data demonstrate that third-party BK-VST infusions are safe and potentially effective, supporting a scalable immunotherapeutic approach for this otherwise fatal disease. (ClinicalTrials.gov NCT02479698).