The study, conducted by Özge Sönmez and Ahmet Emre Eşkazan of the Division of Hematology at Cerrahpaşa Faculty of Medicine, Istanbul University-Cerrahpaşa, used a web-based questionnaire to probe how hematologists manage extreme thrombocytosis, the technical term for platelet counts that far exceed the normal range of roughly 150,000 to 450,000 per microliter. Fifty-six hematologists participated in the survey, which was supported by the Turkish Society of Hematology and approved by the local clinical studies ethics committee at Istanbul University-Cerrahpaşa. The authors set out to map real-world clinical behavior in a domain where evidence is thin, expert opinion varies widely, and the consequences of over-treatment or under-treatment can be serious for patients with essential thrombocythemia, one of the classical Philadelphia-negative myeloproliferative neoplasms.

The central finding concerns the platelet threshold at which physicians choose to intensify therapy. Cytoreductive treatment, which uses medications such as hydroxyurea to suppress platelet production in the marrow, was favored by 41.1 percent of respondents once platelet counts reached or exceeded 1000 × 10⁹ per liter, the celebrated million-platelet mark. But when the threshold was raised to 1500 × 10⁹ per liter, the proportion of hematologists opting for cytoreduction jumped to 83.9 percent. In other words, the million mark alone is not a universal trigger for action; it is a point of debate. Only when counts approach one and a half million does a strong majority of specialists converge on drug-based platelet lowering, a pattern that mirrors the cautious, risk-adapted philosophy embedded in contemporary myeloproliferative neoplasm guidelines.

Why does the million mark inspire such hesitation? The answer lies in the biology of extreme thrombocytosis. Very high platelet counts do not simply increase the risk of blood clots; paradoxically, they also raise the risk of bleeding. The mechanism is acquired von Willebrand disease, a condition in which the sheer excess of circulating platelets adsorbs and proteolytically cleaves von Willebrand factor, the adhesive protein that platelets and clotting depend on. As platelets multiply, they effectively devour their own hemostatic infrastructure, and the largest von Willebrand multimers disappear first. A patient with a platelet count of a million or more may therefore present with gastrointestinal bleeding or bruising rather than thrombosis, and giving aspirin to such a patient in the belief that high platelets mean high clot risk can precipitate dangerous hemorrhage.

This biological nuance explains the second pillar of the survey: laboratory assessment. A striking 78.6 percent of hematologists reported testing for von Willebrand factor ristocetin cofactor activity, abbreviated vWF:RCo, at the time of diagnosis of extreme thrombocytosis. This assay measures the functional capacity of von Willebrand factor to support platelet adhesion under shear stress, and it is the workhorse test for detecting the acquired form of the deficiency. The near-universal uptake of this test among survey respondents suggests that the field has internalized the lesson that platelet number alone is an unreliable guide to bleeding risk, and that documenting acquired von Willebrand disease is now considered a standard part of the initial evaluation when counts soar into the extreme range.

Aspirin decisions revealed an equally disciplined logic. In the presence of cardiovascular risk factors and in the absence of acquired von Willebrand disease, 91.1 percent of hematologists favored aspirin therapy for patients with platelet counts of at least 1000 × 10⁹ per liter, and 78.6 percent favored it at counts of at least 1500 × 10⁹ per liter. The slight decline at the higher threshold is telling: the very patients with the most extreme counts are also the most likely to harbor acquired von Willebrand disease, so the antiplatelet drug that protects one patient’s arteries becomes a hemorrhagic hazard for another. The survey thus captures, in numerical form, the delicate balancing act that defines extreme thrombocytosis management, where the same count can argue for and against the same drug depending on a single laboratory result.

When asked why they initiate cytoreductive treatment at all, respondents cited three goals with roughly equal frequency: prevention of thrombosis, prevention of bleeding, and control of symptoms such as headaches, erythromelalgia, and visual disturbances. This even distribution is itself informative. It shows that hematologists do not view cytoreduction solely as an anti-thrombotic strategy. Lowering the platelet count can restore von Willebrand factor levels and relieve bleeding, and it can also ameliorate the microvascular symptoms that plague many patients with essential thrombocythemia regardless of their thrombotic risk score. The therapeutic rationale, in practice, is tripartite, and the survey suggests that experienced clinicians weigh all three aims rather than fixating on any single one.

The most novel element of the report is its comparison with artificial intelligence. The authors posed the same clinical scenarios to AI programs and recorded the recommendations the models produced. The AI-generated advice supported von Willebrand testing, endorsed cytoreductive treatment particularly at platelet counts of 1500 × 10⁹ per liter or above, and generally recommended aspirin only for patients with counts of at least 1000 × 10⁹ per liter who did not have acquired von Willebrand disease. Across several clinical scenarios, the differences between the human specialists and the machine recommendations were few. The authors are careful to frame this comparison as exploratory, and they stop short of claiming that AI can substitute for clinical judgment, but the convergence is noteworthy at a time when large language models are being proposed as clinical decision aids.

The significance of this convergence deserves careful interpretation. On one hand, it suggests that the consensus positions in hematology, such as testing for acquired von Willebrand disease and reserving aggressive cytoreduction for the most extreme counts, are encoded deeply enough in the published literature that AI systems trained on that literature reproduce them faithfully. On the other hand, it raises the question of whether AI is genuinely reasoning or simply echoing the dominant patterns of its training data, including their blind spots. Where expert opinion is divided, as it clearly is at the million-platelet threshold where only about four in ten specialists opt for cytoreduction, an AI that confidently recommends a single course of action could flatten legitimate clinical nuance into false certainty. The survey’s exploratory design is a prudent acknowledgment of that risk.

For patients with essential thrombocythemia, the practical message of the study is reassuring in one respect and cautionary in another. Reassuring, because the management of extreme thrombocytosis appears to follow a coherent, biologically informed playbook: check von Willebrand function, tailor aspirin to its result, and reserve cytoreductive drugs chiefly for the most extreme counts or for patients with thrombosis, bleeding, or disabling symptoms. Cautionary, because the survey also exposes genuine variability at the margins, with more than half of specialists declining to treat at exactly one million platelets while others intervene. As Sönmez and Eşkazan conclude, extreme thrombocytosis itself remains an important driver of treatment decisions, particularly above 1500 × 10⁹ per liter, and the near-agreement between experienced hematologists and AI recommendations offers a provocative glimpse of a future in which algorithms may help standardize care, provided they are validated rigorously and deployed with the same humility that the survey’s authors bring to their own exploratory analysis.

Subject of Research: Clinical management of extreme thrombocytosis in essential thrombocythemia, comparing hematologist survey responses with artificial intelligence recommendations

Article Title: When platelets reach a million: insights from hematologists

Article References: Sönmez, Ö., & Eşkazan, A. E. (2026). When platelets reach a million: insights from hematologists. Annals of Hematology. https://doi.org/10.1007/s00277-026-07183-x

Image Credits: AI Generated

DOI: 10.1007/s00277-026-07183-x

Keywords: essential thrombocythemia, extreme thrombocytosis, platelets, cytoreductive therapy, acquired von Willebrand disease, aspirin, von Willebrand factor, myeloproliferative neoplasms, hematology, artificial intelligence, clinical survey, thrombosis

The myeloproliferative neoplasms encompass essential thrombocythaemia, polycythaemia vera, prefibrotic primary myelofibrosis, and overt primary myelofibrosis, all disorders driven by mutated haematopoietic stem cells, most commonly carrying the JAK2V617F mutation. These conditions predispose patients to life-threatening blood clots and bleeding, and conventional cytoreductive therapy has primarily aimed at reducing this thrombotic risk. Hydroxyurea has long served as the default first-line option, yet it leaves 10 to 40 percent of patients intolerant or resistant, and lingering concerns about its long-term leukemogenic potential have fuelled the search for alternatives.

Interferon alpha occupies a fundamentally different therapeutic niche. Rather than simply poisoning dividing cells, this immunomodulating agent acts directly on haematopoietic stem and progenitor cells. Animal studies have shown that interferon alpha rouses quiescent malignant stem cells from dormancy, with a striking preference for cells carrying the JAK2V617F mutation, ultimately driving the malignant clone toward functional exhaustion. This disease-modifying potential, combined with the improved tolerability conferred by pegylation, prompted Danish investigators to design DALIAH as the earliest and largest randomised trial of long-term pegylated interferon in newly diagnosed patients across all myeloproliferative neoplasm subtypes.

Between February 2012 and July 2015, the trial enrolled 206 patients, of whom 203 formed the modified intention-to-treat population. Adults with newly diagnosed or treatment-naive disease were eligible regardless of risk score, a deliberate design choice that broadened the trial beyond the high-risk patients targeted by most prior studies. Older patients, aged over sixty, were randomised among hydroxyurea, pegylated interferon alpha-2a, or pegylated interferon alpha-2b, while younger patients avoided hydroxyurea owing to its theoretical leukemogenic risk and were randomised between the two interferon formulations. The cohort comprised 73 patients with essential thrombocythaemia, 89 with polycythaemia vera, 16 with prefibrotic myelofibrosis, and 25 with overt primary myelofibrosis. The median age was 62 years, and JAK2V617F was the dominant driver mutation, present in 74 percent of participants with a median baseline variant allele frequency of 34 percent.

The primary endpoint was the rate of molecular response, defined by European LeukemiaNet criteria as a substantial or complete reduction in the JAK2V617F variant allele frequency, measured serially by quantitative polymerase chain reaction at eight time points up to 60 months. By intention-to-treat analysis, which counted patients who discontinued treatment as non-responders, molecular response rates were statistically indistinguishable between the two drugs: 23 percent in the hydroxyurea arm versus 24 percent in the interferon arm at 60 months. This apparent parity, however, masks a profound divergence revealed when the analysis was restricted to patients who remained on their assigned therapy.

Among those who stayed the course, pegylated interferon proved decisively superior from 36 months onward. At that juncture, 56 percent of interferon-treated patients had achieved a molecular response, compared with just 23 percent of those on hydroxyurea, a gap that widened to 67 percent versus 35 percent by 60 months. Three interferon-treated patients reached complete molecular remission, with the JAK2V617F mutation rendered undetectable by an assay sensitive to 0.1 percent. Moreover, patients on hydroxyurea lost their responses far more frequently, with 62 percent losing molecular response versus only 13 percent of interferon patients, a difference that proved highly significant. The kinetics told a parallel story: the median relative reduction in variant allele frequency from baseline to 60 months was 75 percent with interferon compared with only 20 percent under hydroxyurea.

On secondary endpoints, the two drugs traded advantages. Complete clinicohaematological response at 12 months was similar between groups, and by intention-to-treat analysis hydroxyurea held a modest edge at 18 months, 58 percent versus 38 percent, though per-protocol analysis later favoured interferon at 36 and 60 months. Hydroxyurea normalised blood counts faster, with haematological response reached in a median of 1.6 months versus 3.8 months for interferon, a predictable consequence of its direct cytorepressive action. Conversely, more hydroxyurea patients lost their haematological responses over time, 86 percent compared with 44 percent on interferon. Bone marrow histopathology delivered an unexpected finding: by intention-to-treat analysis, histopathological remission at 60 months favoured hydroxyurea, 18 percent versus 5 percent, though paired analyses and per-protocol comparisons blurred this difference, and the investigators caution that fibrosis grading was inconclusive in significantly more hydroxyurea patients.

Safety data underscored the trial’s central challenge. At 60 months, 60 percent of all patients had discontinued study treatment, and discontinuation was significantly more frequent with interferon at 65 percent than with hydroxyurea at 37 percent. Toxicity drove the disparity: 45 percent of interferon patients stopped treatment for adverse events, most commonly flu-like illness, injection-site irritation, fatigue, and neuropsychiatric symptoms, while only 13 percent of hydroxyurea patients discontinued for that reason. Notably, discontinuation for toxicity clustered among younger and female patients and varied dramatically between study sites, from 30 to 69 percent, leading the authors to speculate that investigator enthusiasm may have shaped dosing decisions. Yet a subgroup of patients tolerated interferon long-term, and no treatment-related events of grade 3 or higher occurred beyond 24 months among those continuing therapy.

Thrombosis, the cardinal clinical concern in these diseases, was not prevented by interferon. Nineteen major thrombotic events occurred during treatment, with a rate of 5.5 events per 100 patient-years among older interferon patients versus 2.8 per 100 patient-years with hydroxyurea, a difference lacking statistical significance but clinically noteworthy. Importantly, 68 percent of thrombotic events occurred while blood counts remained normalised and before any substantial molecular reduction, reinforcing the multifactorial nature of clotting risk in myeloproliferative neoplasms. No patient progressed to myelofibrosis, acute myeloid leukaemia, or myelodysplastic syndrome, and five deaths were recorded across the cohort without an evident treatment-related pattern.

The DALIAH results arrive amid converging evidence that molecular response is not merely a laboratory curiosity. Data from the CONTINUATION-PV and MAJIC-PV trials link deepening JAK2V617F depletion with prolonged event-free survival, suggesting that eradicating the malignant clone may genuinely alter disease course. Within this framework, interferon’s slow but sustained molecular advantage, even as hydroxyurea’s responses eroded, positions low-dose pegylated interferon alpha-2a as a viable first-line option for carefully selected patients, particularly younger individuals in whom decades of cytoreductive therapy loom ahead. The trial’s authors call for biomarkers capable of predicting both response and tolerability at diagnosis, noting that mutated DNMT3A may confer interferon resistance. They also propose an intriguing strategy: an initial six to twelve months of combination therapy using hydroxyurea to rapidly normalise counts while interferon gradually strips away the malignant clone, followed by interferon monotherapy. With pegylated interferon alpha-2b withdrawn from the European market but alpha-2a still available and mono-pegylated ropeginterferon expanding the armamentarium, the era of disease-modifying first-line therapy for myeloproliferative neoplasms may finally be within reach.

Subject of Research: A five-year randomised phase 3 trial comparing pegylated interferon alpha with hydroxyurea as first-line cytoreductive therapy in myeloproliferative neoplasms.

Article Title: Interferon-α vs hydroxyurea in patients with myeloproliferative neoplasms (DALIAH): a multicentre, randomised, open-label, phase 3 trial in Denmark

Article References: Knudsen, T. A., Hansen, D. L., Ocias, L. F., Bjerrum, O. W., Brabrand, M., Christensen, S. F., Eickhardt-Dalbøge, C. S., Ellervik, C., El Fassi, D., Frederiksen, M., Kjær, L., Kristensen, T. K., Kruse, T. A., Larsen, M. K., Mourits-Andersen, T., Möller, S., Overgaard, U. M., Severinsen, M. T., Skov, V., … Hasselbalch, H. C. (2026). Interferon-α vs hydroxyurea in patients with myeloproliferative neoplasms (DALIAH): a multicentre, randomised, open-label, phase 3 trial in Denmark. eClinicalMedicine, 100, Article 104193. https://doi.org/10.1016/j.eclinm.2026.104193

Image Credits: AI Generated

DOI: 10.1016/j.eclinm.2026.104193

Keywords: myeloproliferative neoplasms, pegylated interferon alpha, hydroxyurea, JAK2V617F, molecular response, DALIAH trial, polycythaemia vera, essential thrombocythaemia, primary myelofibrosis, phase 3 trial, cytoreductive therapy, haematology