External validation of an online predictive model for myelodysplastic syndrome
Bibliographic record
Abstract
Abstract Background: Single-lineage cytopenia is highly prevalent in the elderly population, with common causes including nutritional deficiency, erythropoietin or thrombopoietin deficiency, inflammation, idiopathic or clonal cytopenias of undetermined significance, or alterations in stem cell physiology. Older studies have shown an estimated prevalence of myelodysplastic syndrome (MDS) in 30% of unexplained anemia or cytopenia. While bone marrow examination (BME) is the gold standard for the diagnosis of MDS, it is associated with risks, costs, subjective interpretation and non-diagnostic findings. A simple non-invasive diagnostic algorithm has been developed to guide the pretest probability of MDS, with the potential to obviate unnecessary BME. We performed an external validation in a cohort of patients who underwent BME for cytopenia. Aims: To examine the diagnostic accuracy of the MDS prediction algorithm in unexplained cytopenias. Methods: In this retrospective cohort study, we identified consecutive BMEs evaluated for unexplained cytopenias by the Department of Pathology at the University of Alberta (2022-2023). We included patients >50 years who underwent BME for single or multi-lineage cytopenias. Those with known causes of cytopenia, or high suspicion of hematologic malignancy (e.g. circulating blasts, abnormal peripheral blood immunophenotype) were excluded. We estimated the pretest probability of MDS using the web-based predictive algorithm established by Oster et al, categorized as probable MDS, indeterminate, and probably not MDS. We considered BME findings of MDS and myelodysplastic/myeloproliferative neoplasms (MDS/MPN) as positive cases, and all others to be negative cases. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated. Two hematologists independently reviewed each case to adjudicate whether BME changed management. Institutional Research Ethics Board approval was obtained. Results: Overall, 64 patients who underwent BME for cytopenias were included, with a median age of 71 years, and 42 (66%) male. Indications for BME included: bi- or pancytopenia in 40 (63%), isolated anemia in 19 (30%), isolated neutropenia in 3 (5%), and isolated thrombocytopenia in 2 (3%). Thirteen (20%) patients had additional indications for BME, including adenopathy/splenomegaly in 9 (14%), monocytosis in 5 (8%), constitutional symptoms in 2 (3%), and elevated tryptase in 1 (2%). The diagnostic yield of abnormal BME was 36 (56%), while 18 (28%) were normal/ benign, and 10 (16%) non-diagnostic. Abnormal BME findings included MDS or MDS/MPN (26; 41%), lymphoid neoplasms (4; 6%), clonal hematopoiesis (3; 5%), acute leukemia (2; 3%), and metastatic cancer (1; 2%). The model predicted probable MDS in 25 (39%), probably not MDS in 23 (36%), and indeterminate in 16 (25%). Sensitivity, specificity (excluding indeterminate cases), PPV, and NPV were 77%, 61%, 52%, and 82%, respectively. While 19/23 (83%) cases of “probably not MDS” were correctly categorized as non-MDS, only 11 (48%) BMEs could have been safely avoided without missed pathologies (i.e. normal BME). The rest included: 4 (17%) misclassified MDS cases, another 4 (17%) non-diagnostic marrows requiring repeat BMEs, 3 (13%) lymphoid neoplasm, 1 (4%) clonal hematopoiesis. The model correctly predicted MDS in 13/25 (52%) “probable MDS” cases, 4 (16%) had another hematologic neoplasm, 4 (16%) non-diagnostic, only 4 (16%) were normal. At last follow-up, 19 (30%) were deceased. Of the 26 MDS cases, 19 (73%) BMEs led to changes in management, including initiation of hypomethylating agents (13; 50%), hydroxyurea (4; 15%), erythropoietin stimulating agents (3; 12%). In contrast, 7/26 (27%) did not undergo changes in management following BME, due to lower-risk MDS or unfit for chemotherapy. Conclusions: We found that nearly a third of BMEs ordered for unexplained cytopenia had low pretest probability for MDS when applying the MDS risk prediction model. Of these, half could have been obviated without missing any pathologies. We confirmed a high NPV in keeping with prior reports. PPV and specificity were lower in our study, possibly explained by non-diagnostic samples and acute leukemia classified in the “non-MDS” group. Prospective studies are needed to evaluate the role of this model in aiding clinical decision-making.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.013 | 0.031 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".