Development of a generalizable UMAP-based approach for comparing clinical flow cytometry data with application to NPM1-mutated AML cohorts

Introduction

AML with mutated NPM1 is associated with heterogeneous clinicopathologic features. We sought to study the association between phenotype, genetics, and clinical behavior using treatment-naïve bone marrow samples of NPM1-mutated AML. Prior phenotypic studies using flow cytometry data have primarily focused on the blast population in isolation or used less comprehensive analysis techniques, such as simple visual histogram assessment. We applied a dimensionality-reduction algorithm (UMAP) to analyze retrospective clinical flow cytometry data of tumor and non-tumor cells in an initial small cohort of NPM1-mutated samples.

Methods

Custom software was developed using python, FlowKit, and umap-learn to create a dictionary of the various antibody panels, detect the antibody panels that were used in each raw data (FCS) file, and determine the flow cytometer channels that should be disregarded. Subsamples from each FCS file for a given antibody panel were combined and analyzed using UMAP to create an embedding that could then be applied to all FCS files of the given antibody panel. FCS files were subsequently prepared for analysis in FlowJo, including using UMAP coordinates. The initial pilot phase included analysis of the EuroFlow AML1 panel of 11 cases, which included 3 primary refractory cases, 3 early relapsed cases, and 5 cases which achieved clinical remission without relapse. FlowJo was used to gate and examine the clusters identified by UMAP with respect to phenotypic parameters. These same UMAP gates were applied to all 11 cases for direct comparison.

Results

The blast count ranged from 50 to 88 in these cases. The blast phenotype was determined to be myeloid (n=3), monocytic (n=4), or other (CD34-/HLA-DR-)(n=4). Although standard CD45 by side scatter gating delineates four major cell types (lymphocytes, monocytes, granulocytes, blasts), gating using the UMAP algorithm with input data from eight phenotypic markers in conjunction with scattering parameters, produced at least 10 distinct UMAP clusters of variable cellular composition. Interestingly, applying those gates to side scatter (SSC-A) by CD45 histoplots revealed 5 distinct gates falling into the traditional “blast” region of the histoplot. The UMAP gates thus identified provided quantitative values for further statistical analysis.

Conclusion

We have developed a useful tool to automatically identify the antibody panels used to generate prior flow cytometry data, preprocess the data, and apply the UMAP algorithm for creating embeddings that can be applied to additional cases. Our preliminary analyses revealed significant phenotypic heterogeneity among a small cohort of NPM1-mutated cases. Ongoing work includes expansion of the cohort and number of antibody panels incorporated into the analyses to elucidate prognostic and predictive features of tumor and non-tumor populations in treatment naïve samples of NPM1-mutated AML.

Comparison of Multiple Clinical Testing Modalities for Assessment of NPM1-Mutant AML

Background

NPM1 mutation status can influence prognosis and management in AML. Accordingly, clinical testing (i.e., RT-PCR, NGS and IHC) for mutant NPM1 is increasing in order to detect residual disease in AML, alongside flow cytometry (FC). However, the relationship of the results from RT-PCR to traditional NGS, IHC and FC is not widely known among many practitioners. Herein, we aim to: i) describe the performance of RT-PCR compared to traditional NGS and IHC for the detection of mutant NPM1 in clinical practice, and also compare it to FC, and ii) provide our observations regarding the advantages and disadvantages of each approach in order to inform future clinical testing algorithms.

Methods

Peripheral blood and bone marrow samples collected for clinical testing at variable time points during patient management were tested by quantitative, real-time, RT-PCR and results were compared to findings from a Myeloid NGS panel, mutant NPM1 IHC and FC.

Results

RT-PCR showed superior sensitivity compared to NGS, IHC and FC with the main challenge of NGS, IHC and FC being the ability to identify a low disease burden (<0.5% NCN by RT-PCR). Nevertheless, the positive predictive value of NGS, IHC and FC were each ≥ 80% indicating that positive results by those assays are typically associated with RT-PCR positivity. IHC, unlike bulk methods (RT-PCR, NGS and FC), is able provide information regarding cellular/architectural context of disease in biopsies. FC did not identify any NPM1-mutated residual disease not already detected by RT-PCR, NGS or IHC.

Conclusion

Overall, our findings demonstrate that RT-PCR shows superior sensitivity compared to a traditional Myeloid NGS, suggesting the need for “deep-sequencing” NGS panels for NGS-based monitoring of residual disease in NPM1-mutant AML. IHC provides complementary cytomorphologic information to RT-PCR. Lastly, FC may not be necessary in the setting of post-therapy follow up for NPM1-mutated AML. Together, these findings can help inform future clinical testing algorithms.

Automated Calculation of Turnaround Times Using Data Warehouses for Publication in Online Test Menus

Awareness of turnaround times (TATs) is important for time-sensitive clinical decision making and setting expectations for timing communications with patients. In fact, a significant number of clinicians consider the TAT as the most important measure of laboratory quality (PMID 24839360). TAT definitions vary considerably, ranging from rough estimates to more precisely defined statistics based on historical data, the former more typically provided in online test catalogs and the latter usually reserved for a limited number of tests for internal quality assurance and improvement. As data warehousing of laboratory information is becoming more commonplace, calculation of TATs can more easily be performed. Our goal in this project was to develop a generalized, automated approach to calculate expected TATs for publication in our online test catalog. We began by defining TAT as the amount of time between receipt of the specimen by the laboratory and the reporting of the finalized result in the laboratory information system (using specimen collection time increased median TATs by 21 minutes). The TAT of an order set/panel of tests was calculated using the time stamp of the last finalized test result of the set. For development purposes, a subset of warehouse data corresponding to all tests within a 1-month period was selected (170,363 accession numbers, 2,206,311 total test results). All accessions with “add-on requests” (tests requested by physicians using specimens collected for prior tests) were removed. The median and 90th percentile TATs were calculated using R for each test and panel, as these values are commonly used in quality assurance programs and are thought to be intuitively easy to understand in a test menu. Values were not reported if fewer than 100 test results were available for a given test. A summary table and TAT distribution graphs are provided for review by laboratory personnel before being published in the online test menu. Categorizing tests ordered “stat” versus “nonstat” generated some interesting insights, the most obvious being that, for the majority of our tests, the difference in median TATs is very small. In some cases, however, the median stat TAT was actually longer (eg, hemoglobin A1C), which might be due to blood draws at irregular hours by phlebotomists, then batching of samples with specimens collected later at regular blood draw times, suggesting that elimination of the stat option was appropriate. Distinct advantages of using data warehouses to calculate TATs include being able to provide clinicians with data-driven estimates of expected lab result TATs. Such data can also be used for internal quality assurance purposes. Areas of further development include developing a more rigorous mathematical theory of the precision associated with our median and 90th percentile TATs and tuning the frequency of data query and TAT recalculation.

Modulation of an optical needle's reflectivity alters the average photon path through scattering media

We introduce the concept of deliberate placement of absorbers to alter the average path of photons through tissue for a biomedical optical device. By changing the reflectivity of a needle that separates a source and detector, the average photon path through a turbid medium can be changed. Totally reflective needles have photon scattering density functions similar to a point source and detector in an infinite medium. An absorbing needle moves the average photon path of photons that reach the detector away from the needle. Thus, by modulating the reflectivity of the needle, it is possible to modify the sensitive volume, and simple tomography data should be possible. These results are confirmed by Monte Carlo simulations and experiment. Experiments include moving a black target relative to an optical "needle" and measuring the resulting intensity and phase lag of light reaching a detector at the distal end of the needle.

Cloning and characterization of a complex DNA fingerprinting probe for Candida parapsilosis

Candida parapsilosis accounts for a significant number of nosocomial fungemias, but in fact, no effective and verified genetic fingerprinting method has emerged for assessing the relatedness of independent isolates for epidemiological studies. A complex 15-kb DNA fingerprinting probe, Cp3-13, was therefore isolated from a library of C. parapsilosis genomic DNA fragments. The efficacy of Cp3-13 for DNA fingerprinting was verified by a comparison of its clustering capacity with those of randomly amplified polymorphic DNA analysis and internally transcribed spacer region sequencing, by testing species specificity, and by assessing its capacity to identify microevolutionary changes both in vitro and in vivo. Southern blot hybridization of EcoRI/SalI-digested DNA with Cp3-13 provides a fingerprinting system that (i) identifies the same strain in independent isolates, (ii) discriminates between unrelated isolates, (iii) separates independent isolates into valid groups in a dendrogram, (iv) identifies microevolution in infecting populations, and (v) is amenable to automatic computer-assisted DNA fingerprint analysis. This probe is now available for epidemiological studies.