Naomi B. Haas
VerifiedUniversity of Pennsylvania · Rehabilitation Medicine
Active 1984–2026
About
Naomi B. Haas, MD, is a Professor of Medicine (Hematology-Oncology) at the Hospital of the University of Pennsylvania. She is the Director of the Prostate and Kidney Cancer Program at the University of Pennsylvania and serves as a Co-Leader of the Cancer Therapeutics Program at the Abramson Cancer Center. Dr. Haas is recognized as an international expert in the conduct and design of adjuvant clinical trials for kidney cancer and is a national expert in prostate and kidney cancer therapeutics. Her clinical and research focus is on developing and evaluating therapeutic strategies for prostate and kidney cancers, contributing significantly to the advancement of cancer therapeutics through her leadership and expertise.
Research topics
- Internal medicine
- Medicine
- Cancer research
- Oncology
- Pathology
- Surgery
- Immunology
Selected publications
OncoEducate: A pilot study of generative AI to support patient education in GU cancer care.
Journal of Clinical Oncology · 2026-03-01
article713 Background: Genitourinary (GU) cancer care involves complex treatment plans and goals, which can be difficult for patients to understand and navigate. Limited understanding can lead to distress and avoidable care utilization. We developed OncoEducate, a generative artificial intelligence (AI) tool that produces patient-friendly handouts, and piloted its accuracy, feasibility, and acceptability. Methods: OncoEducate uses the Claude Opus 4.1 large language model. User inputs include patient’s diagnosis, treatment intent, and treatment regimen. Based on these inputs, the tool generates tailored, two-page handouts summarizing diagnosis, treatment intent, regimen details, side effects, and reasons to contact the care team. Information is drawn from reputable public sources and standardized language is used for key concepts (e.g., treatment intent). Feedback from patient advocates, oncologists, and pharmacists informed iterative refinement. We conducted a prospective two-phase pilot study (IRB exempt, University of Pennsylvania). Phase 1: Nine handouts for common palliative-intent regimens in advanced kidney, prostate, and urothelial cancers were generated and reviewed by five GU oncologists and an advanced practice provider for accuracy, completeness, and readability. Feedback informed prompt refinement, yielding Version 2. Phase 2: Patients initiating these regimens received Version 2 handouts. Surveys assessed usefulness, readability, and acceptability (7-point Likert scale items) and comprehension of treatment intent (validated item from Cancer Care Outcomes Research and Surveillance study). Analyses were descriptive. Results: Clinicians rated Version 1 handouts as highly accurate and appropriate (median 6-7/7 across domains). Over eight weeks, 20 out of 21 eligible patients enrolled and completed surveys. Most were male (n = 17, 85%) with urothelial (n = 11, 55%) or kidney (n = 6, 30%) cancer. Median age was 71 (range: 46-83). Enfortumab vedotin + pembrolizumab (n = 8, 40%) was the most frequent regimen. Patients strongly agreed that the handouts were informative, easy to read, and improved their understanding of treatment plans and care team contact (median 7/7 for all). Comfort with AI-assisted education was high (median 7/7, range 2-7). 65% (n = 13) of patients correctly identified their treatment intent as palliative - exceeding historical benchmarks (19-31%, Weeks et al, NEJM 2012). Conclusions: OncoEducate handouts were well received by clinicians and patients and may improve understanding of treatment intent. Despite small sample size, findings demonstrate the feasibility of generative AI to deliver concise, personalized education. Larger randomized studies are needed to assess impact on patient outcomes.
The Oncologist · 2026-04-07
articleOpen accessOncoEducate is a clinician-supervised generative artificial intelligence (AI) application designed to deliver standardized, regimen-level, plain-language education at the point of care. We developed a templated handout generator using a locked prompt and predefined content structure that incorporates diagnosis, treatment regimen, and treatment intent, with clinician review prior to patient distribution. We then conducted a prospective, two-phase pilot at a single academic cancer center. In Phase I, six clinicians evaluated AI-generated handouts for nine commonly used palliative-intent genitourinary (GU) oncology regimens using 7-point Likert-scale items. In Phase II, patients with advanced kidney, prostate, or urothelial cancers initiating palliative-intent therapy received clinician-reviewed handouts and completed follow-up surveys at the next office or infusion visit or by telephone when in-person completion was not feasible. Clinicians rated handouts as useful (median 6/7) and accessible (median 6.5/7), with median accuracy ratings of 6-7/7 across domains. During an 8-week period, 20 of 21 approached patients enrolled (95%). Patients rated handouts as informative and readable (median 7/7 for both), and 65% correctly identified treatment intent as palliative. These findings support the feasibility and acceptability of clinician-reviewed, templated AI-generated handouts in routine GU oncology care, justifying larger randomized studies to evaluate clinical impact.
Journal of Clinical Oncology · 2026-03-01
articleTPS580 Background: Ipilimumab plus nivolumab (ipi/nivo) is a standard of care for treatment-naïve, metastatic clear cell renal cell carcinoma (ccRCC). While durability of response may be superior to combination therapy with PD-1 inhibitors and agent tyrosine kinase inhibitors, ipi/nivo has a relatively low objective response rate (ORR) and relatively high progressive disease (PD) as best response. Building on the benefits of ipi/nivo through additional immune manipulation has biologic rationale and could expand the efficacy of this regimen. Lymphocyte-associated gene 3 (LAG3) is an inhibitory receptor on activated immune cells that plays a role in T cell exhaustion. LAG3 is expressed on T cells and is co-expressed with other inhibitory receptors such as PD-1. This co-expression induces the disfunction of T cells and thus limits the anti-tumor T cell response. Relatlimab is a human LAG-3-specific antibody that binds to the LAG-3 receptor with high affinity and blocks LAG-3 interactions with its canonical ligand, major histocompatibility complex (MHC) Class II, which is the peptide antigen presentation molecule recognized by CD4+ T cells. Relatlimab binding inhibits the negative regulatory function of LAG-3 in vitro. Relatlimab restores effector function of exhausted T cells. Combined LAG3 inhibition with relatlimab and PD-1 inhibition with nivolumab was recently demonstrated in the RELATIVITY-047 melanoma study. These compelling preclinical and clinical data have led to the hypothesis that adding relatlimab to ipilimumab plus nivolumab will increase objective response rate in patients with metastatic ccRCC. Methods: This is a phase 2 randomized, multicenter, study investigating the efficacy of nivolumab 480 mg every 4 weeks (Q4W), relatlimab 160 mg Q4W and ipilimumab 1 mg/kg every 8 weeks (Q8W) intravenous (IV) versus a doublet arm treating with nivolumab 480 mg Q3W and ipilimumab 1mg/kg Q3W IV followed by maintenance nivolumab in first-line advanced ccRCC. Co-primary endpoints include safety and ORR. Up to 60 patients with treatment naive, biopsy-proven ccRCC with adequate organ/marrow function with at least one evaluable lesion by RECIST 1.1 will be enrolled and randomized 2:1 to the experimental arm versus ipi/nivo. Ongoing safety and futility monitoring will employ a 2-arm Bayesian optimal phase 2 (BOP2) design. Sample size provides 64% power at a one-sided 0.15 significance level to detect a difference between the arms assuming the overall response rates are 40% vs. 60% for nivo/ipi vs. experimental arm, respectively. Secondary endpoints include progression free survival and OS. To explore the effects of the treatment on inducing activated T cell infiltration, biopsies and circulating immune cell subsets will be obtained for single cell analysis and spatial transcriptomic studies. Over 10 patients are enrolled at the time of submission. Clinical trial information: NCT06708949 .
Clinical Cancer Research · 2026-03-18
articlePURPOSE: We report the long-term clinical outcomes of a multicenter, randomized phase II trial (NCT00089193) that tested immunogenicity of a vaccine composed of 12 class I MHC-restricted melanoma peptides (12MP), with or without granulocyte-macrophage colony-stimulating factor (GM-CSF) as adjuvant and administered at one or two sites in patients with resected high-risk melanoma. PATIENTS AND METHODS: Participants were randomized to one of four treatment arms: 12MP at one site (Arm A), 12MP+GM-CSF at one site (Arm B), 12MP at two sites (Arm C), 12MP+GM-CSF at two sites (Arm D). The trial was powered to detect differences in immunogenicity by vaccine groups defined by GM-CSF status (Arms B+D vs A+C) and vaccine sites (Arms A+B vs C+D). For this analysis, overall survival (OS) and recurrence-free survival (RFS) were evaluated by these vaccine groups. RESULTS: All eligible participants (n=121) were evaluated. Median follow-up was 5.6 years. No significant differences in RFS or OS were found by GM-CSF status. Participants vaccinated at two sites compared to one had significantly improved RFS (HR 0.59, 95%CI: 0.38-0.93, p=0.02) and a trend to improved OS (HR 0.64, 95%CI: 0.39-1.06, p=0.08). On landmark multivariable analysis, two-site vaccination was the only significant predictor of RFS (HR 0.55, 95%CI: 0.34-0.88, p=0.01) after adjusting for CD8+ T cell response and other prognostic factors. CONCLUSIONS: These results challenge the use of GM-CSF as a local vaccine adjuvant and support two-site vaccination. Future work to characterize the locoregional immune response to cancer vaccination at the injection site and vaccine-draining lymph nodes is warranted.
Journal of Clinical Oncology · 2026-03-01
article52 Background: Cardiovascular (CV) disease is a major cause of morbidity and mortality among patients (pts) with prostate cancer (PCa). Shared epidemiologic and androgen-deprivation therapy (ADT)-related factors may contribute to increased CV risk in pts with PCa. As there are limited clinically available tools to educate pts with PCa on CV health risks and CV risk mitigation strategies, CV risk factors are under-recognized and under-treated. We conducted a pilot study to evaluate the feasibility, data quality, and clinical impact of a patient-oriented, web-based application to directly inform PCa pts of their estimated CV risk and CV mitigation strategies (NCT06064149). Methods: Between 11/2023 and 4/2025, PCa patients (< 75 years) currently receiving or planned for ≥ 6 months ADT-based PCa therapy were enrolled. Pts were sent a website link for the CARE-PC application via the electronic medical record (EMR), and self-entered demographic, CV risk factor, CV disease and PCa medical history. Upon completion, CARE-PC immediately provided a personalized CV risk estimate and targeted, guideline-based CV risk reducing education. Primary endpoint was the application completion rate. Secondary endpoints included accuracy of patient-entered data (compared to EMR standard), prevalence of CV risk factors and disease, and cardiologist care access. Results: 82 pts met eligibility. The median age was 66 years, and 20% of pts had localized PCa and 80% had either biochemically recurrent or metastatic PCa. The CARE-PC application completion rate was 54% (44/82). Pts were more likely to complete the web-based tool if age ≥ 65 yrs (61% vs 42%), married (57% vs 46%), Caucasian (62% vs 14% African American), and recurrent or metastatic PCa (59% vs 31% localized PCa). Among 44 pts completing the web tool, 7% (3/44) of pts had diabetes and 61% (27/44) were using antihypertensives (both reported with a 98% (43/44) accuracy rate). 41 pts (93%) accurately reported their CV disease history; 16% (7/44) of pts had a confirmed history of ≥ 1 CV disease. 52% (23/44) of pts were currently followed by a cardiologist. 86% (38/44) accurately reported PCa clinical status (localized vs recurrent or metastatic). 70% (31/44) and 73% (32/44) of pts accurately reported prior PCa therapies and current PCa therapies, respectively. Conclusions: The CARE-PC web-based tool is feasible, but completion rates vary based on age, race, and marital status. Self-entered accuracy of CV risk factor and CV disease was high, with lower accuracy for prior/current PCa therapy. Ongoing analyses will evaluate pt engagement in CV care following CARE-PC participation. Further provider- and patient-oriented quality improvement initiatives are needed to increase awareness and mitigation of competing CV health risks within an individualized PCa clinical context. Clinical trial information: NCT06064149 .
The Oncologist · 2026-02-07
articleOpen accessBACKGROUND: The GRade, Age, Nodes and Tumor (GRANT) score is one of the prognostic models recommended by international guidelines to refine recurrence risk stratification in patients with surgically treated renal cell carcinoma (RCC) and integrates age, tumor size, grade, and nodal status. In this study, we aimed to validate the GRANT score within the ECOG-ACRIN EA8143 PROSPER prospective trial. METHODS: We conducted a validation analysis of the GRANT score within the phase III randomized EA8143 PROSPER study of perioperative nivolumab in surgically treated RCC. Patients were classified into 2 risk groups, favorable (0-1 risk factors) versus unfavorable (2-4 risk factors). Relapse-free survival (RFS) and overall survival (OS) were estimated using the Kaplan-Meier method. Model discrimination were evaluated using Harrell's C-index. RESULTS: Among 714 patients included, 58.3% were favorable and 41.7% unfavorable based on the GRANT score. Patients in the favorable group had a significantly longer median RFS (61.1 vs. 36.9 months; hazard ratio [HR]: 0.36, 95% confidence interval [CI]: 0.27-0.48, P < .001) and OS (median not reached, HR: 0.25, 95% CI: 0.15-0.42, P < .001) compared to patients in the unfavorable group. The c-index was 0.63 and 0.66 for RFS and OS, respectively. A better prognostic performance was observed among nonclear cell RCC for both RFS (HR: 0.13, 95% CI: 0.05-0.33, P < .001; c-index: 0.74) and OS (HR: 0.14, 95% CI: 0.04-0.50, P < .001; c-index 0.74). CONCLUSIONS: The GRANT score was prospectively validated in the PROSPER study, demonstrating prognostic value for both RFS and OS, especially in nonclear RCC, further supporting its use in clinical practice. Clinical trial registration number: ClinicalTrials.gov, NCT03055013.
The Oncologist · 2025-10-01
articleOpen accessAbstract Background Tumor-associated macrophage (TAM) infiltration has been shown to modulate response to immune checkpoint inhibitors in various cancers, but its role in metastatic clear cell renal cell carcinoma (mccRCC) remains unclear. Here, we investigated the role of CD163+ TAMs as a potential determinant of clinical outcomes to first-line anti-PD-1 therapy (nivolumab) in patients with mccRCC enrolled in the HCRN GU16-260 trial. Moreover, as recent data suggest that the interaction between TAMs and tumor infiltrating lymphocytes (TILs) promotes T cell exhaustion, we explored the spatial relationship between CD163+ TAMs and CD8+ TILs in different states of exhaustion (ie, terminally exhausted (TE) and non-terminally exhausted (NTE) CD8+ TILs). Methods Pre-treatment tumor samples from 67 patients were analyzed by multiplex immunofluorescence to identify CD163+ TAMs, CD8+PD-1+TIM-3+ and/or LAG-3 + (TE CD8+), and CD8+PD-1+TIM-3−LAG-3− (NTE CD8+) TILs. Associations between the natural log of density of CD163+ TAMs with progression-free survival (PFS) and objective response rate (ORR) were assessed using univariable Cox and logistic regression models, respectively. An optimized cutoff was determined using minimum p value for ORR. For each tumor, the density of TE CD8+ TILs and the density of NTE CD8+ TILs were calculated within a 30 µm radius area centered on CD163+ TAMs (proximal area) and outside of this area (non-proximal area), using the ‘sf’ package within R software. The densities of TE and NTE CD8+ TILs were compared in proximal versus non-proximal areas across all tumor samples using the Wilcoxon signed-rank test. For each CD8+ TIL population (TE and NTE), the enrichment in proximity of CD163+ TAMs was assessed by calculating the difference in densities in proximal and non-proximal areas normalized by the density in the overall tumor area. The level of enrichment in TE CD8+ TILs versus NTE CD8+ TILs in proximity of CD163+ TAMs was compared using the Wilcoxon signed-rank test. Results The density of CD163+ TAMs, analyzed as a continuous variable was positively associated with ORR (OR: 2.21, 95% CI: 1.33 to 3.69, P = .002) and PFS (HR: 0.77, 95% CI: 0.61 to 0.97, P = .028). At an optimized cutoff, patients with high density of CD163+ TAMs (n = 34, 50.7%) had higher ORR (65% vs. 15%, P &lt; .001) and longer median PFS (16.6 months, 95% CI: 5.5-32.9 vs. 5.5 months, 95% CI: 4.1-10.6, P = .009) compared to patients with low density of CD163+ TAMs (n = 33, 49.3%). The density of CD163+ TAMs was moderately correlated with the density of TE CD8+ TILs (Spearman correlation, r = 0.55) and weakly correlated with the density of NTE CD8+ TILs (r = 0.32). Proximity analysis showed that the density of TE CD8+ TILs was significantly higher in the area proximal to the CD163+ TAMs compared to the non-proximal area (median density: 123.3/mm2 vs. 37.2/mm2; P &lt; .001). Similarly, the density of NTE CD8+ TILs was significantly higher in the area proximal to the CD163+ TAMs compared to the non-proximal area (median density: 127.2/mm2 vs. 66.8/mm2; P &lt; .001). The level of enrichment in proximity of CD163+ TAMs was higher for TE CD8+ TILs compared to NTE CD8+ TILs (0.77 vs. 0.58; p = 0.0011). Conclusions High levels of CD163+ TAMs are associated with improved outcomes to anti-PD-1 therapy in mccRCC. In addition, exhausted CD8+ TILs preferentially localize in proximity of CD163+ TAMs in ccRCC tissues, supporting that TAM-T cell interactions are critical for driving T cell dysfunction. Taken together, our data are consistent with the hypothesis that the efficacy of PD-1 blockade may be in part mediated by reprogramming TAMs from a pro-tumorigenic to an anti-tumorigenic state.
2025-12-11
articleOpen access<p>Marker gene expression in CAR T-cell clusters. A, Uniform manifold approximation and projection (UMAP) plot of AAVS1 and EGR2 knockout (KO) CAR T-cell samples is shown. B, UMAP plots showing expression levels of CD4 and CD8. C, Violin plot depicting expression of cluster-defining markers in CD4+ T-cells. D, Differentially expressed genes in IL7R+ versus CTLA4+ CD4+ T-cells. E, Violin plots showing expression levels of cluster-defining markers in CD8+ T-cells. F, Heatmap displaying differentially expressed genes between CD8+ cell clusters. G, Cell cycle scores mapped on UMAP plots.</p>
2025-12-11
articleOpen access<p>Gene expression and pathway enrichment analysis of CD8+ T cell clusters. A, Heatmap showing differentially expressed genes between memory-like KLF2+ and exhausted-like MKI67+ CD8+ T-cells. Gene signature scores related to cell cycle and clinical response are indicated on the top bars. B, Top downregulated GO biological processes in EGR2 compared to AAVS1 knockout CAR T-cells.</p>
CA A Cancer Journal for Clinicians · 2025-12-13
articleOpen accessThe therapeutic landscape for renal cell carcinoma (RCC) and urinary tract cancer (UTC) has transformed dramatically, creating complexity in treatment selection and sequencing. The 2025 Advanced Urologic Cancer Consensus Conference was convened to establish evidence-based expert consensus recommendations for optimal management. A multidisciplinary panel of 51 experts participated in a modified Delphi process addressing questions developed through iterative consensus-building covering RCC and UTC management. Voting occurred before and after the conference, and analyses focused on postmeeting responses. Consensus was defined as ≥75% agreement, with strong consensus as >90%. Strong consensus was found on the use of adjuvant pembrolizumab for higher risk RCC (pathologic T2 [pT2], grade 4; pT3-pT4, any grade; pTXN1; or fully resected metastatic disease) and on neoadjuvant therapy before cystectomy for localized UTC. There was strong consensus on the use of enfortumab vedotin plus pembrolizumab as frontline therapy for metastatic UTC and the use of platinum-based chemotherapy postprogression in biomarker-negative UTC. For RCC, there was consensus on the role of single-agent vascular endothelial growth factor receptor-tyrosine kinase inhibitor therapy after progression on frontline immune checkpoint inhibitor/vascular endothelial growth factor receptor-tyrosine kinase inhibitor therapy or dual immune checkpoint inhibitor therapy. However, there was a lack of consensus on other critical areas in the management of RCC and UTC. The 2025 Advanced Urologic Cancer Consensus Conference provides evidence-informed guidance for complex clinical scenarios while identifying critical research priorities. The group recognizes that the lack of consensus across multiple areas highlights the need for improved patient selection and prospective studies enabling optimal combination and sequencing approaches. This iterative annual process will address evolving treatment paradigms to optimize outcomes.
Recent grants
Radiobiology and Imaging Program
NIH · $93.0M · 1997–2027
NIH · $328k · 2000
Frequent coauthors
- 268 shared
Toni K. Choueiri
Dana-Farber Cancer Institute
- 212 shared
Vivek Narayan
University of Pennsylvania
- 191 shared
David F. McDermott
- 190 shared
Jae‐Lyun Lee
Ulsan College
- 146 shared
Se Hoon Park
Samsung Medical Center
- 142 shared
Marine Gross‐Goupil
Bordeaux Population Health
- 140 shared
Balaji Venugopal
Beatson West of Scotland Cancer Centre
- 137 shared
Hans J. Hammers
Southwestern Medical Center
Labs
Naomi B. Haas LabPI
- Resume-aware match score
- Save to shortlist
- AI-drafted outreach
See your match with Naomi B. Haas
PhdFit ranks faculty by your research interests, methods, and publications — grounded in their actual work, not templates.
- Free to start
- No credit card
- 30-second signup