To the Editor,
1.
Providers often face the dilemma of determining which patients with erythrocytosis warrant testing for a JAK2 mutation. Only 1%–5% of cases are driven by JAK2 [1], but these patients are critical to identify because JAK2 raises concern for polycythemia vera (PV), which is associated with increased risk of stroke, myocardial infarction, and thrombosis when left untreated [2, 3]. The JAKPOT rule, developed at a tertiary care center in Canada, was designed to guide providers in this challenging scenario [4]. For JAKPOT, patients with zero of three laboratory criteria (red blood cell count > 6.45 × 1012/L, platelets > 350 × 109/L, neutrophils > 6.2 × 109/L) can forego JAK2 testing. Validation on an internal cohort revealed a sensitivity and negative predictive value (NPV) of 100%. We externally validate JAKPOT in three United States (US) cohorts, then execute a quantitative decision analytic model to assess which strategy is cost‐effective in the US healthcare system: JAKPOT versus upfront JAK2 testing for all individuals with erythrocytosis.
We identified all adult patients who had JAK2 testing at Beth Israel Deaconess Medical Center (BIDMC) in Massachusetts; University of Illinois College of Medicine Peoria (UIC) in Illinois; and the Veterans Affairs (VA) health system. We extracted age, sex, race/ethnicity, date and result of JAK2 testing, hemoglobin level, and the JAKPOT laboratory variables. Only patients with elevated hemoglobin (≥ 16.0 g/dL for females, ≥ 16.5 g/dL for males) were included, in line with the criteria used to develop the initial JAKPOT rule. We also only included patients with all three JAKPOT variables available on the same day as JAK2 testing, to limit confounders. Patients with preexisting myeloid neoplasms prior to JAK2 testing were excluded. The JAKPOT rule was applied to all patients who met inclusion criteria, and sensitivity, specificity, and NPV with 95% confidence intervals (CI) to predict positive JAK2 were calculated for each cohort.
We developed a Markov cohort model to evaluate the cost‐effectiveness of the JAKPOT prediction rule (“JAKPOT strategy”) versus status quo (“upfront JAK2 testing”) for individuals with erythrocytosis at median age 60, with a male to female ratio of 1.6 [5]. In the JAKPOT strategy, only individuals with positive JAKPOT undergo JAK2 testing. If JAK2 is detected, they are diagnosed with PV. Those with negative JAKPOT do not proceed with JAK2 testing, resulting in a subset that includes individuals correctly ruled out for PV and those with a missed diagnosis. We presumed JAKPOT negative patients with persistent erythrocytosis would be clinically reevaluated at 1 year (base‐case) or sooner, at 6 months (scenario analyses). In the status quo strategy, all individuals with erythrocytosis undergo JAK2 testing. We assumed perfect sensitivity and specificity of JAK2 mutation testing since it is the current gold standard. Patients diagnosed with PV may receive treatment with aspirin, hydroxyurea, and phlebotomy. Individuals with a missed diagnosis of PV are not treated, with outcomes informed by international, prospective PV natural history data across 4393 person‐years [6], (1) randomized trial data on treated (Hematocrit < 45%) versus untreated (Hematocrit > 45%) PV patients [7], (2) PV‐specific health resource utilization and utilities [8, 9], and (3) long‐term impacts on quality of life and health resource utilization for PV patients with thrombotic or nonfatal cardiovascular events (ischemic stroke, myocardial infarction, venous thromboembolism) [10, 11, 12, 13]. The details of the Markov model and input parameters are provided in Supporting Informations and Table S1.
We designated BIDMC and the VA as dual base‐case analyses and included UIC as a scenario analysis. The prevalence of PV and test characteristics of JAKPOT were informed by each cohort. Our analysis was conducted from the US health system perspective over a lifetime time‐horizon, discounting costs and outcomes at 3% annually. The primary outcome was the incremental cost‐effectiveness ratio (ICER) and its reformulation, the incremental net monetary benefit (iNMB) at a willingness‐to‐pay (WTP) threshold of $100 000/QALY [14]. The secondary outcomes were threshold analyses for the minimum JAKPOT rule sensitivity and JAK2‐positive prevalence that would allow cost‐effective application of JAKPOT. We performed deterministic and probabilistic sensitivity analyses (PSA) to evaluate the effect of all parameters on model outcome (Supporting Informations). TreeAge Pro Healthcare 2024 (TreeAge Software, Williamstown, MA) was used to build the model and conduct analyses. The Consolidated Health Economic Evaluation Reporting Standards (CHEERS) reporting guideline was implemented where applicable (Supporting Information).
We identified 2308 patients at BIDMC (2013–2023), 1819 patients at UIC (2015–2024), and 50 827 patients at the VA (2006–2023) who had JAK2 testing. After exclusion criteria, there were 476, 681, and 8288 patients remaining for JAKPOT validation, respectively (Figure S1). The median age ranged from 58 to 65 years and all three cohorts were majority male (73.1%–97.2%) and white (67.6%–88.4%) (Table 1). In the BIDMC, UIC, and VA cohorts, the sensitivity of JAKPOT was 0.94 (95% CI: 0.86–0.98), 0.73 (0.61–0.82), and 0.81 (0.77–0.84); specificity was 0.52 (0.47–0.57), 0.69 (0.65–0.73), and 0.69 (0.68–0.70); and NPV was 0.98 (0.96–0.99), 0.95 (0.93–0.97), and 0.98 (0.977–0.984), respectively (Table S2).
TABLE 1.
Patient demographics of validation cohorts.
| Characteristics | BIDMC (n = 476) | UIC (n = 681) | VA (n = 8288) |
|---|---|---|---|
| Median age (range) | 65 (21–89) | 58 (18–92) | 63 (19–101) |
| Male sex | 348 (73.1%) | 512 (75.2%) | 8054 (97.2%) |
| Race/Ethnicity | |||
| White | 322 (67.6%) | 602 (88.4%) | 6638 (80.1%) |
| Black | 36 (7.6%) | 40 (5.9%) | 558 (6.7%) |
| Asian | 32 (6.7%) | 9 (1.3%) | 57 (0.7%) |
| American Indian | 2 (0.4%) | 4 (0.6%) | 40 (0.5%) |
| Hispanic/Latino | 19 (4.0%) | 16 (2.3%) | 346 (4.2%) |
| Other/Unknown | 65 (13.7%) | 10 (1.5%) | 649 (7.8%) |
| JAK2 positivity rate | 70 (14.7%) | 73 (10.7%) | 549 (6.6%) |
| Any JAK2, CALR, or MPL positivity rate | 71 (14.9%) | 73 (10.7%) | 549 (6.6%) |
Abbreviations: BIDMC, Beth Israel Deaconess Medical Center; UIC, University of Illinois College of Medicine Peoria; VA, Veterans Affairs.
In the dual base‐case within the BIDMC and VA cohorts, the status quo strategy of upfront JAK2 testing accrued 360 and 510 more population lifetime QALYs at a cost of $6.9 and $30.3 million more than the JAKPOT strategy, respectively. These results yielded ICERs of $19 000/QALY [95% credible interval (CI) $3100–$55 000/QALY] and $60 000/QALY [95% CI: $41 000–$88 000/QALY], below the US $100 000/QALY willingness to pay threshold, and positive iNMBs of $29.4 million [95% CI: $7.5–69.5 million] and $20.5 million [95% CI: $4.5–$40.6 million], thus favoring upfront JAK2 testing (Table 2). The results of the UIC cohort and the 6‐month re‐evaluation scenario analyses are shown in Table S3.
TABLE 2.
Population level base case results*.
| Test strategy | Cost ($) | Incremental cost ($) | Effectiveness (QALYs) | Incremental effectiveness (QALYs) | Incremental cost‐effectiveness ratio ($/QALY) [95% CI] | iNMB at WTP of $100 000 per QALY ($) [95% CI] | % cost‐effective at WTP of $50 000|$100 000|$150 000 per QALY |
|---|---|---|---|---|---|---|---|
| Base case (BIDMC cohort) | |||||||
| JAKPOT | 2.20 billion | 1.10 million | — | — | 4|0|0 | ||
| Upfront JAK2 testing | 2.21 billion | 6.9 million | 1.10 million | 360 | 19 000 [3100–55 000] | 29.4 million [7.5–69.5 million] | 96|100|100 |
| Base case (VA cohort) | |||||||
| JAKPOT | 1.00 billion | 1.12 million | — | — | 82|0|0 | ||
| Upfront JAK2 testing | 1.03 billion | 30.3 million | 1.12 million | 510 | 60 000 [41 000–88 000] | 20.5 million [4.5–40.6 million] | 18|100|100 |
Population referred to a cohort of 60‐year‐old Americans with erythrocytosis (~88 000 individuals).
No parameter variation ±20% of the absolute point estimate changed the conclusion that upfront JAK2 testing is the cost‐effective strategy (Figure S2). The minimum test sensitivities of JAKPOT required for it to be cost effective at different times to patient re‐evaluation are shown in Figure 1. For JAKPOT to be cost effective, the minimum prevalence of JAK2‐positivity in a tertiary referral center clinical context (BIDMC or UIC) or a VA‐wide clinical context was 4.5% and 4.2%, respectively. In PSA in both cohorts, upfront JAK2 testing was favored in (1) 96%, 100%, and 100% and (2) 18%, 100%, and 100% of 10 000 Monte Carlo iterations at WTP thresholds of $50 000/QALY, $100 000/QALY and $150 000/QALY, respectively (Table 2).
FIGURE 1.
Two‐way sensitivity analysis of time to re‐evaluation versus JAKPOT sensitivity.
In this analysis, JAKPOT did not meet the minimum sensitivity threshold for cost‐effectiveness (0.984), with observed sensitivities of 0.94 (BIDMC), 0.73 (UIC), and 0.81 (VA). The JAKPOT test would be cost‐effective in clinical settings (1) that can re‐evaluate patients within 6 weeks or (2) that have a JAK2 prevalence rate below 4.2%. Reducing time to follow‐up represents a practical opportunity for health systems to improve diagnostic safety and cost‐efficiency, rather than relying solely on improved risk models. The risk of missing PV is also lower in populations with lower JAK2 prevalence. While tertiary cohorts (including the original JAKPOT derivation cohort at 11.7% [4], BIDMC at 14.7%, and UIC at 10.7%) show higher positivity rates, prevalence is likely lower in primary care settings, where JAKPOT may be better suited.
Our study is strengthened by the size of our multi‐institutional external validation, with nearly 10 000 US patients, and model parameterization with the largest prospective PV natural history data and randomized controlled trial data. Our study also has limitations. First, our model does not consider the development of myelofibrosis or acute leukemia, which are major causes of death for patients with PV. However, because earlier intervention with aspirin, phlebotomy, or hydroxyurea treatment is not thought to change the risk of disease progression, the natural history of PV would be the same in both strategies and therefore unlikely to impact our findings. As evidence emerges for whether treatments such as pegylated interferon can affect allele frequency and the course of disease, these interventions will need to be accounted for in future models [15]. Second, while our model included all types of JAK2 testing (including V617F quantification and exon 12‐15), we could not identify the specific type of JAK2 test used in our study cohorts. Test sensitivity for detecting JAK2 can vary, but restricting our cohort to patients with elevated hemoglobin likely selected for clinically relevant disease rather than JAK2 CHIP.
In conclusion, the JAKPOT rule effectively identifies which patients with erythrocytosis are unlikely to have a JAK2 mutation in a large US cohort. However, due to insufficient test sensitivity, upfront JAK2 testing is the cost‐effective approach in the US. Health systems can assess and tailor the cost‐effectiveness of JAKPOT by considering practical factors such as local follow‐up capacity and JAK2 mutation prevalence. This versatility allows for a dynamic, locally adaptable approach to diagnostic stewardship, making our findings relevant across a range of care settings.
Author Contributions
B.D.L., S.I., J.R., R.P., J.L., J.A.F., and G.G. conceived the study design. All authors wrote and edited the manuscript.
Funding
The authors have nothing to report.
Ethics Statement
This research was approved by the institutional review boards (IRB) at Beth Israel Deaconess Medical Center (#2023P001136) and University of Illinois College of Medicine at Peoria (#2204697‐1) and by the VA Boston Healthcare System Research and Development Committee as an exempt study prior to data collection and analysis, with a waiver of informed consent per the Common Rule due to the use of existing data.
Conflicts of Interest
R.P. has served as a consultant for Merck Research and Sanofi and has received travel funding from Octapharma; none of these activities were related to this study. All other authors declare no conflicts of interest.
Supporting information
Data S1: CHEERS Checklist.
Table S1: Model input parameters.
Table S2: JAKPOT external validation results.
Table S3: Scenario analyses: UIC cohort, BIDMC and VA cohorts (6‐month re‐evaluation).
Figure S1: Flowchart describing inclusion and exclusion criteria applied to cohorts.
Figure S2: Tornado diagrams for the dual base‐case.
Acknowledgments
G.G. is supported by the Frederick A. DeLuca Foundation, Yale Cancer Center, Yale Bunker Endowment; as well as the National Institutes of Health (K01 HL175220) for research unrelated to this work. The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the funding sources.
Data Availability Statement
Data used in this study are from publicly sourced research, outlined within the article, and are available on email request from the corresponding author.
References
- 1. Wouters H. J. C. M., Mulder R., van Zeventer I. A., et al., “Erythrocytosis in the General Population: Clinical Characteristics and Association With Clonal Hematopoiesis,” Blood Advances 4, no. 24 (2020): 6353–6363. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Landolfi R., Marchioli R., Kutti J., et al., “Efficacy and Safety of Low‐Dose Aspirin in Polycythemia Vera,” New England Journal of Medicine 350, no. 2 (2004): 114–124. [DOI] [PubMed] [Google Scholar]
- 3. Barbui T., Finazzi G., and Falanga A., “Myeloproliferative Neoplasms and Thrombosis,” Blood 122, no. 13 (2013): 2176–2184. [DOI] [PubMed] [Google Scholar]
- 4. Chin‐Yee B., Bhai P., Cheong I., et al., “A Rational Approach to JAK2 Mutation Testing in Patients With Elevated Hemoglobin: Results From the JAK2 Prediction Cohort (JAKPOT) Study,” Journal of General Internal Medicine 38, no. 8 (2023): 1828–1833. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Palandri F., Mora B., Gangat N., and Catani L., “Is There a Gender Effect in Polycythemia Vera?,” Annals of Hematology 100, no. 1 (2021): 11–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Marchioli R., Finazzi G., Landolfi R., et al., “Vascular and Neoplastic Risk in a Large Cohort of Patients With Polycythemia Vera,” Journal of Clinical Oncology 23, no. 10 (2005): 2224–2232. [DOI] [PubMed] [Google Scholar]
- 7. Marchioli R., Finazzi G., Specchia G., et al., “Cardiovascular Events and Intensity of Treatment in Polycythemia Vera,” New England Journal of Medicine 368, no. 1 (2013): 22–33. [DOI] [PubMed] [Google Scholar]
- 8. Yu J., Nelson J., Marlin T., Braunstein E., and Jerry M., “Direct and Indirect Costs for Patients With Myeloproliferative Neoplasms,” Leukemia and Lymphoma 65, no. 8 (2024): 1153–1160. [DOI] [PubMed] [Google Scholar]
- 9. Network, G.B.o.D.C , Global Burden of Disease Study 2019 (GBD 2019) Disability Weights (Institute for Health Metrics and Evaluation (IHME), 2020). [Google Scholar]
- 10. Zhu J., Kamel H., Gupta A., et al., “Prioritizing Quality Measures in Acute Stroke Care: A Cost‐Effectiveness Analysis,” Annals of Internal Medicine 176, no. 5 (2023): 649–657. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Schleinitz M. D. and Heidenreich P. A., “A Cost‐Effectiveness Analysis of Combination Antiplatelet Therapy for High‐Risk Acute Coronary Syndromes: Clopidogrel Plus Aspirin Versus Aspirin Alone,” Annals of Internal Medicine 142, no. 4 (2005): 251–259. [DOI] [PubMed] [Google Scholar]
- 12. Parasuraman S. V., Shi N., Paranagama D. C., and Bonafede M., “Health Care Costs and Thromboembolic Events in Hydroxyurea‐Treated Patients With Polycythemia Vera,” Journal of Managed Care and Specialty Pharmacy 24, no. 1 (2018): 47–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Guy H., Laskier V., Fisher M., et al., “Cost‐Effectiveness of Betrixaban Compared With Enoxaparin for Venous Thromboembolism Prophylaxis in Nonsurgical Patients With Acute Medical Illness in the United States,” PharmacoEconomics 37, no. 5 (2019): 701–714. [DOI] [PubMed] [Google Scholar]
- 14. Vanness D. J., Lomas J., and Ahn H., “A Health Opportunity Cost Threshold for Cost‐Effectiveness Analysis in the United States,” Annals of Internal Medicine 174, no. 1 (2021): 25–32. [DOI] [PubMed] [Google Scholar]
- 15. Quintás‐Cardama A., Kantarjian H., Manshouri T., et al., “Pegylated Interferon Alfa‐2a Yields High Rates of Hematologic and Molecular Response in Patients With Advanced Essential Thrombocythemia and Polycythemia Vera,” Journal of Clinical Oncology 27, no. 32 (2009): 5418–5424. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1: CHEERS Checklist.
Table S1: Model input parameters.
Table S2: JAKPOT external validation results.
Table S3: Scenario analyses: UIC cohort, BIDMC and VA cohorts (6‐month re‐evaluation).
Figure S1: Flowchart describing inclusion and exclusion criteria applied to cohorts.
Figure S2: Tornado diagrams for the dual base‐case.
Data Availability Statement
Data used in this study are from publicly sourced research, outlined within the article, and are available on email request from the corresponding author.