Perioperative outcomes in patients with myeloproliferative neoplasms: a multicentric analysis of 354 surgical procedures

TO THE EDITOR:

Polycythemia vera (PV), essential thrombocythemia (ET), and myelofibrosis (MF) are chronic myeloproliferative neoplasms (MPN) whose clinical course is punctuated by risks of both thrombotic and bleeding complications. Although surgery is an established situational risk factor for thrombosis in the general population,¹ this risk is further increased in patients with MPN.² In the perioperative setting, the effects of JAK2V617F mutations, elevated hematocrit (Hct), hyperviscosity, and stasis may be amplified, contributing to higher rates of cardiovascular events.³ In addition, concurrent acquired von Willebrand syndrome and inherent or therapeutic platelet (PLT) dysfunction may increase the risk of hemorrhage. One of the most comprehensive studies on this subject to date corroborated high rates of both perioperative thrombosis (7%) and hemorrhage (10.5%) in patients with PV/ET, despite most (74%) having optimally controlled blood counts.² Importantly, there is currently no consensus or guidelines for perioperative management of MPN, and expert recommendations are based on limited data.^4,5 Correspondingly, practices are heterogeneous and potentially inappropriate, as demonstrated by a recent pan-Canadian study.⁶ Moreover, there are particularly scarce data on MF cohorts and the impact of JAK (Janus kinase) inhibitors, now widespread in this population. This study sought to comparatively assess 90-day perioperative complication rates, risk variables impacting outcomes, and management strategies and their ramifications in a large MPN population, with the goal of informing clinical practice and improving patient outcomes.

This study was approved by institutional review boards and written informed patient consent was obtained. Patients diagnosed with PV, ET, and MF according to World Health Organization criteria⁷ between August 1981 and October 2021, and enrolled in the Quebec MPN Research Group registry (6 academic and community centers) were included. Consecutive cases where the patient had undergone at least 1 surgical procedure since diagnosis, with available pre- and postoperative data, were analyzed. Data were abstracted on demographics, cardiovascular risk factors, thrombosis/hemorrhage history, laboratory values at diagnosis and pre- and postsurgery, type of surgery, therapy, and perioperative modifications. End points included surgical (per procedure) and 90-day postsurgery hemorrhage, arterial and venous thrombosis, and mortality. Major thrombotic and hemorrhagic events were defined per convention.^8,9 Standard surgical definitions were used to categorize interventions as major (general, orthopedic, cardiovascular, and neurosurgery)^10,11; all others were defined as minor. Conventional statistical methods were used (JMP Pro 14.1.0 software; SAS Institute, Cary, NC), with P <.05 considered significant.

A total of 354 procedures were performed in 184 patients: PV, n = 87 (47%); ET, n = 66 (36%); and MF, n = 31 (17%). Demographic and clinical variables at diagnosis are presented in Table 1. The median age at diagnosis was 64 years (range, 19-89 years); 48% male; 82% JAK2V617F mutated (98% PV vs 65% ET/74% MF; P < .0001). At least 1 cardiovascular risk factor was identified in 62% of all patients, and 19% had a history of arterial thrombosis before or at diagnosis, balanced across MPN subtypes. A higher distribution of venous thrombosis before or at diagnosis was found in patients with PV vs ET/MF (13% vs 6%/0%, respectively; P = .04).

The variables at the time of surgery are detailed in Table 2. Of the 354 interventions, major surgeries were performed in 77 cases (22%), primarily in patients with PV vs ET/MF (30% vs 14%/18%, respectively; P = .004; Table 2), and most were general (71% vs 22% orthopedic/6% cardiovascular). Surgeries were urgent in 80 cases (23%), and 79 (40%) were performed under general anesthesia. Blood counts at the time of surgery disclosed significantly higher white blood cell counts (WBC) and lower hemoglobin (Hb) values and PLT counts in MF cases (WBC, 11.1 × 10⁹/L; Hb, 106 g/L; PLT, 239 × 10⁹/L) vs PV (WBC, 8.4 × 10⁹/L; Hb, 136 g/L; PLT, 265 × 10⁹/L) and ET (WBC, 6.3 × 10⁹/L; Hb, 127 g/L; PLT, 365 × 10⁹/L) (all P < .0001). Notably, 101 of 122 evaluable patients with PV (82.8%) had optimally controlled Hct <45% at time of or before surgery. Ongoing therapy at the time of intervention consisted of antiplatelet agents in 255 (78%), cytoreduction in 264 (79%), and anticoagulation in 29 (10%); fewer patients with MF were treated with the aforementioned agents (P < .0001-.005). Cytoreductive agents included hydroxyurea in 177 (67%), ruxolitinib in 45 (17%), anagrelide in 15 (6%), busulfan in 9 (3%), and combinations in 17 (6%). Ruxolitinib was used exclusively in those with MF (n = 35 [63%]) and PV (n = 10 [10%]). When querying surgery-driven modifications of therapy, it was found that antiplatelet therapy was stopped in 38% and 40% of PV and ET cases, respectively, and cytoreduction was suspended in 20% and 11% of PV and ET cases, respectively. Thromboprophylaxis anticoagulation was administered in 42 patients (23%); agents used/doses are summarized in supplemental Figure 1.

Clinical outcomes showed surgical bleeding in 8 patients (2%). At the 90-day follow-up, arterial/venous thrombosis occurred each in 3 patients (1%), hemorrhage in 21 patients (6%), and death in 4 patients (1%). A summary of the individual patient-level data for cases of hemorrhage is presented in supplemental Table 1. The overall complication rate (composite of thrombosis/bleeding and mortality end points) was 11% (n = 38). Patients with PV displayed significantly higher rates of surgical bleeding (6%; P = .006); the remaining complications were evenly distributed among the MPN subtypes.

A separate subanalysis of strictly elective procedures was performed (supplemental Table 2). This disclosed patients with MF to have higher 90-day postoperative hemorrhage as well as total bleeding rates (including surgery; P = .03 and .008, respectively), and a tendency toward higher overall perioperative complication rates (composite end point; P = .05) compared with PV/ET.

Factors impacting 90-day surgical outcomes, including hemorrhage-free survival (HFS) and complication-free survival (CFS) as a composite endpoint, were assessed. Results for HFS were as follows: PV: leukocytosis ≥11 × 10⁹/L (P = .05), low Hct (<35%; P = .03), and antithrombotic prophylaxis (P = .004) negatively impacted HFS on multivariate testing (Table 3). ET: driver mutation status CALR (calreticulin) vs JAK2 (P = .003), PLT count (P = .05), and urgent interventions (P = .007) emerged as independent predictors of HFS. MF: modification or discontinuation of cytoreduction perioperatively was the sole factor impacting HFS (P = .02). Notably, only 3 patients in the overall MPN cohort had PLT count >1 million at the time of surgery (all undergoing urgent/semiurgent procedures precluding optimal count control); whereas none exhibited bleeding complications, cases were too few to permit analysis of the impact of extreme thrombocytosis on outcomes. Variables impacting perioperative complication-free survival were as follows: PV: although several factors were significant on univariate testing, multivariate testing disclosed that only lower Hct (<35%; P = .04) and postoperative thromboprophylaxis (P = .03) had an impact on complication rates. ET: urgent (P < .0001) and major procedures (P = .009), PLT count >450 × 10⁹/L (P = .02), absence of routine antiplatelet therapy preoperatively (P = .02), and postoperative antithrombotic prophylaxis (P = .001) were significant independent predictors of CFS. MF: ruxolitinib exposure (P = .02) was the only factor negatively impacting CFS on the multivariate analysis. When the analysis was restricted to patients who had undergone multiple procedures, history of a previous perioperative complication was a significant risk factor for future complications in patients with ET (P = .04) and MF (P = .003) (trend for PV; P = .07). Limited informative events for thrombosis and survival precluded these analyses. Deaths within 90 days of surgery were reported to be due to underlying pathology or cancer and unrelated to thrombotic, bleeding, or technique-related issues.

Surgical interventions in MPN have been scarcely studied, despite the highly prevalent clinical scenarios faced by clinicians. To our knowledge, this is one of the largest and most contemporary analyses of perioperative outcomes in patients with MPN. Although subject to limitations, including those inherent to the retrospective design, incompletely informative data sets, a wide period of observation, and restricted sample size, several novel and clinically relevant observations were made. Firstly, this study confirms prevalent bleeding perioperatively, although balanced by relatively fewer thrombotic events, calling for further study of thromboprophylaxis and antiplatelet practices in this setting. Second, it discloses key management patterns, notably routine discontinuation of cytoreduction, which appeared to have a detrimental impact on outcomes, although additional corroborative studies are warranted. Likewise, the potential harm of exemption from routine antiplatelet therapy in the preoperative context was observed, consistent with published cardiovascular literature.^12,13 In this series, the latter had an impact on the overall complication rate, which included thrombosis; however, as thrombosis rates were low overall, the underlying explanation for this is not entirely clear and may involve protracted biological mechanisms and/or anti-inflammatory effects of antiplatelet agents. Third, although validation is required, the correlation of uncontrolled PLTs and overzealous Hct control with higher complication rates in ET and PV, respectively, underscores the importance of judicious target control. These are in fact, actionable observations and emphasize supportive data mandating adequate control of PLT count in ET,¹⁴ as well as avoidance and/or correction of anemia in patients with PV before surgery. Furthermore, CALR driver mutation status is associated with bleeding risk in ET, suggesting additional contributing mechanisms conceivably related to acquired von Willebrand and/or other hemostatic defects that might warrant explicit preoperative testing beyond usual standards. Additionally, ruxolitinib use was an independent predictor of perioperative complications in MF, although this may reflect skewing toward higher-risk populations. Importantly, inferior outcomes were recorded in those who had previously experienced surgical complications, denoting them as higher-risk, feasibly helping to inform and optimize the trajectories of future interventions. Finally, although enriched in subgroups having undergone minor procedures, the absence of fatal incidences of thrombosis or hemorrhage in our series provides a measure of reassurance for physicians, highlighting that surgical interventions can generally be safely managed in patients with MPN. Further studies, specified by disease and types of complications, will be required to validate the findings, with these data serving as a sentinel signal for risk awareness and reappraisal of management strategies.

Contribution: N.S., H.J.O., L.B., and S.S. designed, performed research, and collected data; S.D.B., M.T., A.S., S.A., and M.H. provided additional data collection; N.S., H.J.O., and S.S. analyzed the data and wrote the manuscript; and all the authors reviewed and approved the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Natasha Szuber, Department of Hematology, Maisonneuve-Rosemont Hospital, 5415 Blvd de l’Assomption, Montreal, QC H1T 2M4, Canada; email: natasha.szuber@umontreal.ca.