About

John Paul Graff, B.S., D.O., M.A.S., is an assistant professor in the Department of Pathology and Laboratory Medicine at UC Davis Health, specializing in Hematopathology and Informatics. He applies his four board certifications in anatomic pathology, clinical pathology, hematopathology, and clinical informatics to accurately and efficiently diagnose patients, facilitating access to optimal treatments. As an osteopathic trained physician, Dr. Graff considers the whole patient when reviewing biopsies and laboratory data. His clinical interests include malignant and non-malignant hematopathology, with integration of flow cytometry and molecular testing to provide precise diagnoses. In his academic role, he is the associate director of clinical informatics at UC Davis Medical Center, actively training future physicians on how technology can improve patient care. His research involves leveraging machine learning and informatics to classify causes and types of Alzheimer’s dementia, collaborating across UC campuses. Additionally, Dr. Graff has contributed to publications related to COVID-19 and its impacts on the hematologic system, aiding in diagnosis and treatment of critically ill patients.

Research topics

• Computer Science
• Artificial Intelligence
• Machine Learning
• Statistics
• Data Mining
• Mathematics
• Medicine
• Gastroenterology
• Internal medicine

Selected publications

• Casting a wide net for anemia in advanced prostate cancer: A case series

  Current Problems in Cancer Case Reports · 2025-08-18

  articleOpen access

  Anemia is a common occurrence in patients with prostate cancer and has numerous associated etiologies including androgen deprivation, radiation toxicity, bone marrow infiltration from cancer, anemia of chronic inflammation, poor nutritional status and bone marrow suppression from chemotherapy. Anemia in metastatic castration resistant prostate cancer (mCRPC) has been associated with a worse prognosis. While anemia associated with solid tumors generally is attributable to either the malignancy or the treatment given for the malignancy, prostate cancer is unique in that it is an indolent disease with fewer cytotoxic treatments and primarily affecting a geriatric population, thus leading to more varied causes of anemia. Non-treatment related causes of anemia ought to be recognized to prevent further morbidity and mortality in this vulnerable patient population. We outline cases that exemplify this spectrum.

  PublisherDOI

• Diagnosing B-cell acute lymphoblastic leukemia in 2 pediatric patients with recent SARS-CoV-2 infection

  Clinical Pathology · 2024-01-01 · 3 citations

  articleOpen accessSenior authorCorresponding

  COVID-19 infection is still a mystery in terms of its long-term effect on health and its consequences on hematological disorders. Prior studies including ours have shown the abnormal changes in hematopoietic cells in COVID-19 patients. In this article, we are presenting 2 cases of pediatric B-lymphoblastic leukemia (B-ALL) with a previous history of COVID-19 infection. The first case describes a 22-month-old boy presenting with lymphadenopathy, neutropenia, and anemia with concurrent COVID-19 infection without any evidence of a hematolymphoid neoplasm as per bone marrow and lymph node biopsy. However, he presented after 2 months with bone marrow biopsy confirming B-ALL. The second case is that of a 4-year-old girl presenting with B-ALL who has had asymptomatic COVID-19 infection 5 months before this current presentation. Both the cases had complete resolution of COVID-19 infection during the time of presentation with acute leukemia. There were notably 2 rare findings along the course of the patients’ illnesses. First, the unusual plasmacytosis in the marrow during active COVID-19 infection in the first patient and the second, is predilection of development of B-ALL following COVID-19. In both the cases the fluorescence in situ hybridization (FISH) studies showed pathologic alteration of the RUNX1 gene. Overall, there are no literature to support a causal association between acute B-ALL and COVID-19. The diagnosis of B-ALL in these patients after COVID-19 infection may be totally unrelated. However, if we consider Greaves proposed 2-hit model for childhood acute leukemia, that an infectious agent can precipitate development of B-ALL in a genetically susceptible individual. Alteration of the RUNX1 gene in both the patients, opens a door for further exploration of the “second-hit” hypothesis regarding an infectious agent precipitating development of B-ALL in a genetically susceptible individual.

  PublisherDOI

• Learning fast and fine-grained detection of amyloid neuropathologies from coarse-grained expert labels

  bioRxiv (Cold Spring Harbor Laboratory) · 2023-01-17

  preprintOpen access

  Precise, scalable, and quantitative evaluation of whole slide images is crucial in neuropathology. We release a deep learning model for rapid object detection and precise information on the identification, locality, and counts of cored plaques and cerebral amyloid angiopathies (CAAs). We trained this object detector using a repurposed image-tile dataset without any human-drawn bounding boxes. We evaluated the detector on a new manually-annotated dataset of whole slide images (WSIs) from three institutions, four staining procedures, and four human experts. The detector matched the cohort of neuropathology experts, achieving 0.64 (model) vs. 0.64 (cohort) average precision (AP) for cored plaques and 0.75 vs. 0.51 AP for CAAs at a 0.5 IOU threshold. It provided count and locality predictions that correlated with gold-standard CERAD-like WSI scoring (p=0.07± 0.10). The openly-available model can quickly score WSIs in minutes without a GPU on a standard workstation.

  PublisherOA PDFDOI

• Any association between pediatric B-Cell Lymphoblastic Leukemia following SARS-CoV-2 infection: a case series

  2023-06-05

  preprintOpen accessSenior author

  COVID-19 infection is still a mystery in terms of its long-term effect on health and its consequences on hematological disorders. In this article, we are presenting two cases of pediatric B-lymphoblastic leukemia (B-ALL) with a previous history of COVID-19 infection. Notably, in both the cases the fluorescence-in situ hybridization (FISH) studies showed pathologic alteration of the RUNX1 gene. The finding of this genetic alteration along with known pre-leukemic translocation ETV6-RUNX1, opens a door for further exploration of the “second-hit” hypothesis regarding an infectious agent precipitating development of B-ALL in a genetically susceptible individual.

  PublisherOA PDFDOI

• Learning fast and fine-grained detection of amyloid neuropathologies from coarse-grained expert labels

  Communications Biology · 2023-06-24 · 10 citations

  articleOpen access

  Precise, scalable, and quantitative evaluation of whole slide images is crucial in neuropathology. We release a deep learning model for rapid object detection and precise information on the identification, locality, and counts of cored plaques and cerebral amyloid angiopathy (CAA). We trained this object detector using a repurposed image-tile dataset without any human-drawn bounding boxes. We evaluated the detector on a new manually-annotated dataset of whole slide images (WSIs) from three institutions, four staining procedures, and four human experts. The detector matched the cohort of neuropathology experts, achieving 0.64 (model) vs. 0.64 (cohort) average precision (AP) for cored plaques and 0.75 vs. 0.51 AP for CAAs at a 0.5 IOU threshold. It provided count and locality predictions that approximately correlated with gold-standard human CERAD-like WSI scoring (p = 0.07 ± 0.10). The openly-available model can quickly score WSIs in minutes without a GPU on a standard workstation.

  PublisherOA PDFDOI

• Blood Count Parameters As Early Predictors of Clonal Cytopenia of Undetermined Significance (CCUS) Using a Diverse Ethnoracial Data Set

  Blood · 2022-11-15

  article

  Background: Hematologic malignancies are a spectrum of disease that frequently have an insidious onset and may be difficult to diagnose. Improved screening using basic complete blood count parameters are needed to enhance early detection of these processes. Aging results in an expansion of hematopoietic clones that may predispose individuals to subsequent development of a hematologic malignancy. Recent studies with next generation sequencing (NGS) and whole-exome sequencing have identified certain somatic mutations known as DTA mutations that are associated with aging (e.g., DNMT3A, TET2, and ASXL1). These mutations, although more commonly seen in myeloid malignancies, are initially benign but may develop into overt hematologic malignancy and are collectively termed clonal hematopoiesis of indeterminate potential (CHIP). Similarly, individuals with cytopenia and clonal mutations are designated as clonal cytopenia of undetermined significance (CCUS). However, these mutations carry a risk of progression to hematologic malignancy. Animal models have shown that clonal hematopoiesis can accelerate cardiovascular mortality due to increased atherogenesis by causing inflammation and endothelial injury via upregulation of inflammasome pathways. Also, it has already been shown that clonal hematopoiesis with increasing number of clonal mutations, higher variant allele frequency and mutations in known driver genes have a higher risk of progression to hematopoietic malignancy (acute myeloid leukemia, AML). Certain acute phase reactants, like C-reactive protein (CRP) or erythrocyte sedimentation rate (ESR) have been used in prior studies as markers of inflammation and are strongly associated with variations in red cell size as determined by the red cell distribution width (RDW), reflecting chimeric clonal origins. We therefore propose that RDW and mean corpuscular volume (MCV) with or without other blood count parameters (red cell count, white blood cell count and platelet count) might be used as important predictors of progression of myeloid neoplasms, especially myelodysplastic syndromes (MDS), clonal cytopenias (CCUS) and acute myeloid leukemias (AML). Methods: Utilizing Epic's self-service reporting tool SlicerDicer, cohort discovery was performed on the Caboodle Patients data model. Cohort criteria included individuals with the cytopenia diagnostic grouper that had undergone blood or bone marrow myeloid panel next generation sequencing, that also had concurrent or historic complete blood count (CBC) results. Patients with known acute leukemias were excluded. Once the cohort was identified, a data pull was performed that captured CBC values, age, gender, race, and ethnicity, in addition to the NGS results. The NGS results were ranked into 3 classes from 0 to 2. Class 0 included results with no apparent disease, while class 2 results included high grade myeloid mutations (i.e. mutations in NRAS, KRAS, TP53, RUNX1 or ASKL1). Class 1 mutations were considered low grade and featured mutations not seen in class 2. The collated data was analyzed using SAS software. Results: Overall 72 patients met inclusion criteria. Patients ranged in age from 23 to 92 years, 48% identified as female and 52% identified as male with a mixture of race and ethnicity (Table 1). Abnormalities in several CBC components including MCH, MCHC, RBC count, and hemoglobin concentration were associated with increased odds of identifying those with malignant myeloid processes (Figure 1). It is notable that no mutations in FLT3 or KIT were observed in this cohort. Discussion: Based on our preliminary data, these cost-effective parameters can be added to an algorithm for risk assessment of hematologic neoplasms and better predict which patients will further benefit from targeted gene testing. Also, additional data will be helpful and applying this data to modern machine learning techniques will further the clinical application of these findings. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

  PublisherDOI

• Leukoerythroblastosis in a Sickle Cell Patient With Pregnancy: An Interesting Peripheral Blood Smear Finding

  Journal of Hematology · 2022-02-01 · 1 citations

  articleOpen accessSenior author

  The global pandemic of coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has shaken the entire world. The social, health and financial impacts of this pandemic are beyond words. We have learnt a lot about this new disease in a short period of time, but still a long road to go to fully determine its pathogenic effect. The primary target of this virus is angiotensin-converting enzyme 2 (ACE2) receptor, which is prevalent in endothelial cells throughout the body. Immunocompromised patients such as patients with sickle cell disease are more vulnerable to severe respiratory infections, including infection with SARS-CoV-2. In addition, sickle cell disease patients are prone to vaso-occlusive crisis, and theoretically SARS-CoV-2 can worsen the situation as it also can cause endothelial dysfunction and thrombosis. Herein, we are sharing an interesting peripheral blood smear finding of an asymptomatic 31-year-old multigravida pregnant female with a history of sickle cell disease and found to have a positive COVID-19 polymerase chain reaction (PCR) test during her third trimester of pregnancy at a routine clinic visit. Two weeks after the initial positive test, she developed nausea, vomiting, constipation and a pain crisis affecting her extremities while her COVID-19 PCR test was still positive. She was hemodynamically stable, and lab workup revealed chronic anemia, leukocytosis with neutrophilia and lymphopenia. Morphologic examination of the peripheral blood smear showed a marked leukoerythroblastosis: rare myeloblasts, sickle cells, markedly abundant nucleated red blood cells (RBCs), metamyelocytes, and many large and giant platelets were seen. In this context, her previous peripheral blood smears (prior to positive COVID-19 test) did not show leukoerythroblastosis. She was managed conservatively with hydration and pain control and delivered at 36 weeks via cesarean section due to pre-term labor and intrauterine growth retardation. The unusual finding of leukoerythroblastosis in a pregnant sickle cell disease patient with an asymptomatic COVID-19 infection indicates further studies to determine its effect on hematopoietic system and elucidate its clinical significance.

  PublisherOA PDFDOI

• Identifying Multiple Myeloma Patients Using Automated Data Capture from Electronic Medical Records

  Blood · 2022-11-15

  article

  Background: Multiple myeloma (MM) is an incurable plasma cell (PC) malignancy typified by the presence of monoclonal proteins on serum protein electrophoresis (SPEP) and immunofixation or elevation of either kappa or lambda serum free light chains (sFLC) and end organ damage. Electronic Medical Records (EMR) offer a potential source of real world data that is more generalizable than clinical trial data and more granular than registry data. However, MM cohort derivation is challenging: diagnostic codes have poor specificity, and manually created registries are time consuming and frequently focus only on patients diagnosed on site, excluding second opinions or patients transferring care. Attempts to combine diagnostic codes with procedure codes for lab tests, biopsies or radiology studies have reported sensitivities of 73-90% and positive predictive values (PPV) of 60-93%. These approaches relied on test completion, not test results, which may misclassify patients as having MM. As part of an ongoing quality improvement project at the University of California Davis (UCD) designed to identify all patients with MM treated at our institution, we hypothesized that integrating test results, rather than their utilization, with diagnostic codes would create a more specific algorithm while maintaining sensitivity and high PPV. Methods: Patient diagnostic codes, lab, and bone marrow (BM) biopsy reports were loaded from EMR into the Temporal Data Association Platform (TDAP), a patient level data matrix that integrates multi-modal clinical data and organizes it by time allowing for temporal association of clinical events and the ability to assess patient trends over time. Bone marrow (BM) biopsy reports are evaluated by a machine learning algorithm trained to identify the presence and quantity (CD138 percentage) of clonal PCs. Patients identified in the screen set are then passed through both the evaluative and fast track. The UCD algorithm then runs on TDAP and consists of two components (Figure 1): an evaluative track, and a fast-track. Like prior algorithms, the Fast-Track identifies MM specific diagnostic codes, but only evaluates visits at the UCD Cancer Center. The evaluative track classifies patients into categories based the International Myeloma Working Group diagnostic criteria (Figure 2): 1) treatment ineligible smoldering MM (SMM), 2) treatment eligible SMM, 3) symptomatic MM. Patients not meeting criteria for these categories are excluded. Patients enter the UCD MM cohort if identified as treatment eligible SMM or symptomatic MM or via the fast-track. The UCD algorithm was built using a training set of 141 unique patients identified at the UCD Cancer Registry and sampled at random from the EMR. These charts were manually reviewed and labeled as MM, including treatment eligible SMM, (n=112) or not (n=29). To assess the performance of this and a previously described algorithm (Brandenburg 2019) once predictions were made by each algorithm, we assessed accuracy (the proportion of accurate predictions), PPV, and sensitivity. Results: The UCD algorithm identified 111 of the MM cases and excluded MM in 9 of the negative cases with 20 false positives and only 1 false negative. This resulted in a PPV of 85% and sensitivity of 99% with an accuracy of 85%. The evaluative track alone identified 69 (sensitivity 62%) of MM cases and correctly excluded 19 negative cases, for an overall accuracy of 62%. The Brandenburg algorithm identified MM in 67 cases and excluded MM in 28 of the negative cases. It had only 1 false positive case but had 45 false negatives. This resulted in PPV of 99%, but a sensitivity of only 60% for an overall accuracy of 67%. Conclusions: We describe a novel approach to the identification of MM patients incorporating lab and pathology results with diagnostic codes to create a sensitive algorithm while maintaining high accuracy and PPV. Evaluating test results was inadequate, and incorporating diagnostic codes allowed for higher sensitivity. Further refinement to increase specificity while maintaining sensitivity is ongoing. Performance on a validation cohort will be presented at the meeting. By incorporating test results in a temporal framework, TDAP, we can associate temporally related events such as rise in M-spike and new end organ damage. Future directions are expansion of the algorithm to other institutions and automated assessment of MM disease status. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

  PublisherDOI

• Importance of bone marrow clot sections—Positive for metastatic tumor

  International Journal of Laboratory Hematology · 2022-02-16 · 1 citations

  article

  Data available on request from the authors.

  PublisherDOI

• Utility of Peripheral Blood Flow Cytometric Blast Enumeration Reporting in Patients without Prior Hematologic Malignancy

  Blood · 2022-11-15

  article

  Background: The significance of documenting the enumerated data of low levels of blast events (CD34+/CD117+) by peripheral blood flow cytometry studies is largely unaddressed in the literature. The presence of a blast population in peripheral blood of individuals without documented appropriate explanation (such as treatment with a colony-stimulating factor) can be the very first indication of an evolving hematolymphoid neoplastic process. We aim to identify a threshold of quantified blast events on peripheral blood flow cytometric studies that may alert medical professionals of a possible disease process affecting the bone marrow. Methods and materials: Peripheral blood flow cytometry studies performed by the clinical flow cytometry laboratory at the University of California Davis Medical Center between 2015 and 2020 that were detected to have less than or equal to 1% blast population were evaluated. Patients with a history of a known hematologic malignancy at the time of the aforementioned peripheral blood flow cytometry study were excluded. The remaining cases from patients without a known hematological malignancy were investigated. All flow cytometry events were analyzed by multiparametric 10-color flow cytometry using an extensive panel of monoclonal antibodies. The data events were analyzed with Kaluza® software. Percentages of blasts in patients with progression to a hematologic malignancy were compared with the group of patients without progression to a hematologic malignancy. The results of this comparison were utilized to establish a threshold of the presence of CD34+/CD117+ blasts that may warrant formal reporting and recommendations of additional studies. Results: A total of 165 flow cytometry cases from 2015-2020 were evaluated. The patients' electronic medical records were reviewed in order to determine the presence of a prior hematolymphoid malignancy and to evaluate for the progression of a malignancy after the flow study. Of the cases reviewed, 110 were excluded due to the presence of a prior hematolymphoid malignancy. The remaining 55 cases were separated into a group who on follow up had no progression to a hematolymphoid malignancy (n=50), and a second group who developed a hematolymphoid malignancy (n=5) (see figure). The mean percent of CD34+/CD117+ events in the group of patients who progressed to a hematolymphoid malignancy was higher than the non-malignancy control group (0.80% vs 0.40% of total flow cytometry events respectively). The mean percent of CD34+ events in the group of patients who had a malignancy was slightly higher than the control non-malignant group mean percent CD34+ events (0.248% vs 0.220% of total flow cytometry events respectively). Conclusions: We identified a small percentage of patients with no known hematolymphoid malignancy and low levels of blasts on peripheral blood flow cytometry study who later developed a hematolymphoid malignancy (5/55 = 9%). On further investigation, a slightly elevated level of CD34+/CD117+ blast events in these patients was noted. Of the patients who did not develop hematolymphoid malignancies, a significant percentage had non-malignant inflammatory conditions. Based on our current assessment, some patients with low levels of circulating blasts detected by flow cytometry may necessitate follow-ups, including bone marrow evaluation. The presence of cytopenias and/or elevated inflammatory markers may further assist in the decision to proceed with additional hematologic work-up. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

  PublisherDOI

Frequent coauthors

• Ananya Datta Mitra

  University of California Davis Medical Center

  8 shared

• Anupam Mitra

  University of California, Davis

  6 shared

• Denis M. Dwyre

  University of California, Davis

  6 shared

• Hooman H. Rashidi

  University of Pittsburgh Medical Center

  4 shared

• Daniel R. Wong

  Royal Columbian Hospital

  4 shared

• Dennis Lee

  Biological Sciences Curriculum Study

  3 shared

• Kristin Olson

  Minnesota Gastroenterology

  3 shared

• Michael J. Keiser

  3 shared

Awards & honors

• Fellow Teaching Award, UC Davis (2017)