Data Studio

Workshops are an extensive and in-depth consultation for a Medical School researcher based on research questions, data, statistical models, and other material prepared by the researcher with the aid of our facilitator. During the Data Studio Workshop, the researcher explains the project, goals, and needs. Experts in the related topic from across campus will be invited and contribute to the brainstorming. After the meeting, the facilitator will follow up, helping with immediate action items and summary of the discussion. Ultimately, we strive to pair each PI with a data scientist for long-term collaboration.Office Hours are brief consultations for Medical School researchers during the last session of each month. DBDS faculty are available to advise about your research questions. Consult the schedule below to complete the Office Hour registration form. Once you have registered, you will receive a calendar invitation with the date, time, and location of the session. Bring any data, prior analyses, or other materials that you have. Our consultants may even recommend your project for a Workshop if it is appropriate.

One-to-one consultations for Medical School researchers are available year-round. Our facilitator assigns each request to a data scientist with the relevant expertise.

Partners

General questions about statistical issues may be brought to the STAT390 Consulting Workshop. This is a class offered by the Department of Statistics during each academic quarter that is staffed by graduate students and directed by a faculty instructor. The service typically consists of a single meeting with the researcher to address a specific concern, such as planning of experiments and data analysis. For more information, consult the STAT390 Consulting Workshop web page.

Researchers who are members of the Stanford Cancer Institute (SCI) conducting research projects related to cancer may request assistance from the SCI Biostatistics Shared Resource.

The Genetics Bioinformatics Service Center (GBSC) offers an end-to-end bioinformatics consulting service (BaaS) that provides high performance computational infrastructure and cutting-edge bioinformatics services for the Stanford community. The team consults on genomics, transcriptomics, proteomics, epigenetics, and metabolomics projects, and also develop custom workflows. For consulting and hands-on bioinformatics help with your projects please reach out to gbsc-baas-team@lists.stanford.edu to set up an initial meeting.

Schedule

The Data Studio is held each Thursday from 3:00 until 4:30 pm during the fall, winter, and spring quarters of the academic year. Consult the schedule below for the location of each session. Students may participate by enrolling in BMDS 291 for an introduction to the art of statistical consultation and practicum working on projects with a biomedical researcher. All are welcome to attend. Click here to sign up for our mailing list.

The currently scheduled topic is listed below.

 

TITLE: Personalized Circulating Tumor DNA for Measurable Residual Disease Monitoring in AML

SUBTITLE: Designing a Prospective Single-Center Observational Cohort

INVESTIGATORS:

Tian Yi Zhang (1)

Ritika Dutta (1)

Alexandre Wery (1)

Gabriel Mannis (1)

Vanessa Kennedy (2)

Henning Stehr (3)

(1)    Medicine/Hematology

(2)    Medicine/Blood & Marrow Transplantation-Cellular Therapy

(3)    Pathology

DATE: Thursday, 5 November 2026

TIME: 3:00–4:30 PM

LOCATION: Room R358, Edwards Building, 300 Pasteur Drive, Stanford, CA

WEBPAGE: https://dbds.stanford.edu/data-studio/

ABSTRACT

The Data Studio Workshop brings together a biomedical investigator with a group of experts for an in-depth session to solicit advice about statistical and study design issues that arise while planning or conducting a research project. This week, the investigator(s) will discuss the following project with the group.

INTRODUCTION

Acute myeloid leukemia (AML) is an aggressive blood cancer. Most patients treated with chemotherapy reach complete remission (CR), but many later relapse, because leukemia cells persist below the detection limit of routine tests. This residual leukemia is called measurable residual disease (MRD), and its presence after treatment is one of the strongest predictors of relapse and death. Current standard-of-care (SOC) MRD tests have two main limitations. First, they require a bone marrow biopsy. Second, the more sensitive molecular tests (PCR or sequencing of a single mutation such as NPM1, FLT3-ITD, or a core-binding-factor fusion) can only be used in roughly one third of patients. Multiparameter flow cytometry can be used in all patients but is less sensitive.

Patients are routinely grouped into three genetic risk categories by the European LeukemiaNet (ELN) 2022 classification: favorable, intermediate, and adverse. These groups differ substantially in remission rate, relapse rate, treatment (at Stanford, about two thirds of adverse-risk patients receive lower-intensity therapy), and the likelihood of proceeding to allogeneic stem cell transplant (alloHSCT).

Our assay measures leukemia-derived DNA fragments circulating in blood plasma (ctDNA). At diagnosis, we sequence each patient’s leukemia and normal (germline) DNA to find the somatic mutations unique to that patient (a median of about 30 per patient). We then build a personalized capture panel that combines these patient-specific variants with common AML driver mutations. This panel is used to test plasma at each follow-up timepoint. Each sample yields a binary MRD call (detected or not detected) and a quantitative tumor burden (mean tumor allele fraction).

Pilot data (Gunaratne et al., Blood Cancer Discovery 2026) come from a retrospective cohort of 56 patients with a median of 5 sampled timepoints per patient and a median follow-up of 25.8 months. ctDNA-MRD was applicable in 100% of patients, compared with 32% who had a mutation suitable for a SOC molecular MRD test. Among 49 patients in first complete remission (CR1), with 32 relapse-free survival (RFS) events, the hazard ratio (HR) for RFS comparing MRD-positive with MRD-negative patients was 17.8 (95% CI 5.2 to 61.8) for ctDNA, 4.4 (2.1 to 9.2) for combined SOC MRD, and 3.6 (1.7 to 7.5) for flow cytometry. ctDNA was positive before relapse in 28 of 31 patients who relapsed (90%), compared with 16 of 31 (52%) by SOC MRD. Among patients negative by SOC tests, ctDNA turned positive a median of 160 days before clinical relapse. Because the pilot was small and retrospective, the confidence intervals are wide and the effect sizes are likely overestimated.

The proposed study is a single-center, prospective, non-interventional cohort. Adults with newly diagnosed AML (excluding acute promyelocytic leukemia) will be enrolled, and ctDNA results will not be used in clinical care. Patients receive either high-intensity or lower-intensity therapy, and patients from either arm may proceed to alloHSCT. Plasma and blood cells are collected at diagnosis, at the Day 28 response assessment (after induction or cycle 1), after cycle 2, at the end of consolidation in patients not transplanted, every 3 months during observation or ongoing lower-intensity therapy, before alloHSCT, at days +30, +90, and +180 and 1 year after alloHSCT, then every 3 months to 5 years, and at suspected relapse. Bone marrow is collected only from clinically indicated procedures: at diagnosis, after cycle 1 and cycle 2, before alloHSCT, and at the end of treatment, following ELN-DAVID MRD timepoints, and at days +30, +90, and +180 and 1 year after alloHSCT, following Stanford BMT practice. Paired blood cell and bone marrow samples will be analyzed in the first 50 patients.

HYPOTHESIS & AIM

• Research Hypothesis: ctDNA-MRD status at clinically defined landmarks (first remission, pre-transplant, and post-transplant) is associated with subsequent relapse within each ELN 2022 risk group, and predicts relapse more accurately than SOC MRD (flow cytometry and single-mutation molecular assays).
• Project Aim: To determine the enrollment target and analysis framework for the prospective cohort. Specifically, we want to know how many patients are needed overall, within each ELN 2022 risk group, in the transplant subgroup, and within selected genotypes (e.g., TP53-mutated AML), to test the association between ctDNA-MRD and relapse and to define ctDNA thresholds associated with low relapse risk.

DATASET

• Data Type: The planned prospective data include patient-level baseline covariates (age, ELN 2022 risk group, mutation profile, treatment intensity, and alloHSCT with donor type and conditioning intensity). They also include sample-level longitudinal MRD results at irregular, clinically driven timepoints: ctDNA-MRD (binary call, mean tumor allele fraction, and number of variants detected), SOC flow cytometry (binary and percentage), and SOC molecular MRD where available. Event times are recorded for remission, alloHSCT, relapse, and death, with non-relapse mortality treated as a competing event.
• Methodology: Participants will be consented prospectively. Specimens will be processed and sequenced in the Zhang Laboratory, and clinical data will be abstracted from the electronic health record (Epic) into a secure research database (MBioLIMS).
• Existing Data: There are two existing datasets.
• Retrospective pilot cohort (n = 56): median age 60.5 years (range 31 to 85); ELN favorable 25%, intermediate 21%, adverse 54%; 73% received high-intensity induction; 52% underwent alloHSCT; 89% achieved CR; 62% of CR patients relapsed; 55% died; median follow-up 25.8 months; median 5 timepoints per patient (range 2 to 12).
• Zhang Lab institutional biobank of newly diagnosed AML (n = 197): ELN favorable 41 (20.8%), intermediate 60 (30.4%), adverse 96 (48.7%). Lower-intensity therapy was given to 27%, 43%, and 64% of these groups, and alloHSCT to 29%, 60%, and 52%. At a median follow-up of 14 months, crude relapse was 25%, 32%, and 39%, with a median time to relapse from diagnosis of 11, 12.5, and 9 months. The composite CR rate was about 80%.
• Expected accrual: about 60 to 70 newly diagnosed AML patients per year at Stanford, of whom roughly three quarters are expected to be eligible and evaluable, or about 50 patients per year.

STATISTICAL MODELS

Our working sample size calculation is as follows. The primary endpoint is the 2-year cumulative incidence of relapse (CIR) from CR1, comparing ctDNA-MRD-positive with ctDNA-MRD-negative patients. It is analyzed with a Fine–Gray model in which non-relapse mortality is a competing risk, and sized with the Latouche–Porcher (2007) extension of the Schoenfeld formula, using 80% power, two-sided α = 0.05, 2-year non-relapse mortality of 15%, and 10% loss to follow-up. The inputs for each risk group (favorable, intermediate, adverse) are: 2-year CIR 0.25, 0.30, and 0.40; MRD-positive prevalence at CR1 0.35, 0.40, and 0.70; and CR rate 0.95, 0.75, and 0.60.

Assuming a subdistribution HR (sHR) of 5, about 13 to 15 relapse events are needed per risk group. This corresponds to 63, 63, and 67 patients (193 total) if each group is enrolled to its own target. Unselected consecutive enrollment would require about 315 patients, because the favorable group (about 20% of patients) is the limiting group. At an sHR of 3, the requirements rise to 412 (risk-group targets) and 670 (unselected).

Other planned analyses include Kaplan–Meier and Cox models for RFS and overall survival, adjusted for age, ELN risk group, treatment intensity, and transplant status. We also plan cumulative incidence functions and Fine–Gray models for relapse, and landmark analyses at CR1, post-consolidation, pre-transplant, and post-transplant days +30, +90, and +180. Cause-specific Cox models for relapse and non-relapse mortality will complement the Fine–Gray analysis and will be used for time-dependent covariates such as serial ctDNA status and transplant. Diagnostic performance will be assessed by sensitivity, specificity, PPV, NPV, time-dependent ROC/AUC, and C-index, using SOC MRD as the comparator, with SOC flow cytometry MRD classified by the ELN-DAVID 2025 thresholds (positive at 0.1% or higher, low-level positive at 0.01% to below 0.1%). Continuous ctDNA burden will be modeled with restricted cubic splines to explore thresholds. Exploratory analyses include penalized regression and random survival forests. Analyses will be done in R. We are uncertain how best to handle repeated MRD measurements and how to validate ctDNA thresholds within a single cohort.

STATISTICAL QUESTIONS

1. Enrollment by ELN 2022 risk group: Should we enroll each risk group to its own target (about 193 patients total) or enroll all consecutive patients (about 315)? How should the per-group targets be set when CR rate, relapse rate, and MRD-positive prevalence differ across groups, and do group-specific tests need adjustment for multiple comparisons?
2. Primary endpoint and effect size: Is 2-year CIR (Fine–Gray) the right primary endpoint compared with RFS or overall survival? What sHR is reasonable to assume, given that the pilot HRs come from a small retrospective cohort and are Cox hazard ratios rather than subdistribution hazard ratios?
3. Serial measurements: What is the best way to analyze repeated ctDNA measurements taken at irregular, clinically driven timepoints (landmark analyses, time-dependent Cox models, or joint longitudinal and survival models)? How does that choice affect the required sample size?
4. Transplant subgroup: Whether a patient receives alloHSCT is not known at enrollment (29% to 60% depending on risk group). How should we size the pre- and post-transplant MRD analyses, given that time zero is the transplant date and models need to adjust for donor type and conditioning intensity?
5. Genotype-specific thresholds: How many patients and events are needed to define, and then validate, a ctDNA threshold associated with low relapse risk within a specific genotype such as TP53-mutated AML? Should we use separate discovery and validation sets, internal validation such as bootstrapping, or a pooled model with genotype interaction terms?

ZOOM MEETING INFORMATION

Topic: BMDS 291A: Data Studio

Join from PC, Mac, Linux, iOS or Android: https://stanford.zoom.us/j/95373246516?pwd=2X8q3J22JJ5a5s46a7FVSepBz2PBBS.1

    Password: 019839

Or iPhone one-tap (US Toll): +18333021536,,95373246516# or +16507249799,,95373246516#

Or Telephone:

    Dial: +1 650 724 9799 (US, Canada, Caribbean Toll) or +1 833 302 1536 (US, Canada, Caribbean Toll Free)

    Meeting ID: 953 7324 6516

    Password: 019839

    International numbers available: https://stanford.zoom.us/u/awM3GZjwI

    Meeting ID: 953 7324 6516

    Password: 019839

    SIP: 95373246516@zoomcrc.com

    Password: 019839