Pregnancy in patients with myelofibrosis: Mayo–Florence series of 24 pregnancies in 16 women

Among young patients with myeloproliferative neoplasms (MPN), women constitute the majority in essential thrombocythaemia (ET; ˜69% and 64% for ages ≤40 and 41–60 years respectively), whereas the corresponding percentages were lower in polycythaemia vera (PV; 46% and 48%) and primary myelofibrosis (PMF; 41% and 41%).^{1, 2} Accordingly, current literature on pregnancies in MPN is heavily skewed towards ET; in a recent review,³ close to 500 pregnancies in ET were pooled from four independent series with ˜100 or more pregnancies reported in each study, whereas the largest reported experience in PV included 48 women with 121 pregnancies.⁴ Current literature regarding pregnancy outcome in patients with MF is scant with <20 reported cases.^{2, 5-7} In the present study, we describe the experience from two centres of excellence in MPN regarding outcome of pregnancy in MF, including PMF, post-ET MF and pre-fibrotic PMF (pre-PMF).

The present study was conducted as a two-centre series in collaboration with the University of Florence, Italy and the Mayo Clinic, USA after approval by the respective Institutional Review Boards. Study patients fulfilled the 2016 World Health Organization (WHO) and the International Working Group for Myelofibrosis Research and Treatment (IWG-MRT) criteria for the diagnosis of PMF, secondary MF (post-ET/post-PV MF) or pre-PMF respectively.^{8, 9} The 2016 WHO classification emphasises the prognostic relevance of differentiation between ET and pre-PMF and distinguishes pre-PMF from overt PMF, based on the absence of Grade ≥2 bone marrow reticulin fibrosis. To identify informative cases, we utilised the Mayo Clinic and University of Florence MPN databases that included 3023 and 2606 patients respectively, comprising 390 and 416 women aged <50 years; only pregnancies recorded in the pre-PMF, PMF, and MF phase of ET or PV were considered. A comprehensive obstetric history of pregnancies occurring before and after MF diagnosis was recorded. Fetal outcomes were characterised as live birth, first, second trimester loss or stillbirth. Maternal complications included pre-eclampsia, gestational diabetes mellitus, cholestasis, thrombosis and haemorrhage occurring up to 6 weeks postpartum. Statistical analyses considered clinical and laboratory parameters obtained at the time of pregnancy, after diagnosis of MF. Differences in the distribution of continuous variables between categories were analysed by Mann–Whitney test, while nominal variables were compared by chi-squared test. JMP^® Pro 13.0.0 software (SAS Institute, Cary, NC, USA) was used for all statistical analysis.

A total of 24 pregnancies were recorded in 16 women (10 Mayo and six Florence) after diagnosis of MF (PMF, five; post-ET MF, three; pre-PMF, 16); seven women had more than one pregnancy; of which two pregnancies occurred in six women and three pregnancies in only one woman. The median (range) age at initial diagnosis of MF was 29·5 (17–42) years, with a median (range) leucocyte count of 8·1 (2·9–16·7) × 10⁹/l and platelet count of 651 (149−1200) × 10⁹/l. Among 15 women with known driver mutational status, 33% were Janus kinase 2 (JAK2) V617F mutated (five), and the remainder 67% were calreticulin (CALR) mutated; type 1 CALR (seven), type 2 CALR (two) and CALR type unknown (one). Next-generation sequencing was available in 10 patients of which none harboured high-risk mutations, with one patient with zinc finger CCCH-type, RNA binding motif and serine/arginine rich 2 (ZRSR2) and colony-stimulating factor 3 receptor (CSF3R) mutations. Cardiovascular risk factors (hypertension, diabetes mellitus, hyperlipidaemia and smoking) were absent in all cases. Palpable splenomegaly was noted in half of the study patients; ranging from a palpable tip to 11 cm below the left costal margin. Major thrombotic events prior to or at diagnosis occurred in four women (25%); three were splanchnic venous thromboses and one cerebral sinus thrombosis. Major haemorrhage was reported in only one instance. Prior treatments included aspirin and cytoreductive therapy in eight and six women respectively; the latter comprised of hydroxyurea (six), interferon (three), anagrelide (two) and ruxolitinib/pacritinib (one). In addition, three women had received prior anticoagulation. In regard to obstetric history, among seven women with antecedent pregnancies, five (71%) had experienced a prior fetal loss. Of note, none of these fetal losses were in the context of a prior MPN such as ET.

The median (range) age at the time of pregnancy after diagnosis of MF was 33 (25–42) years, with a median (range) leucocyte count of 8·8 (3·5–20) × 10⁹/l and platelet count of 462 (120–1343) × 10⁹/l. Most of these pregnancies were singleton (22 pregnancies). Treatments utilised during pregnancy included aspirin (14; 58%), pegylated interferon (two; 8%) and low-molecular-weight heparin (one; 4%). There were live births in 17 pregnancies (71%), with the remainder experiencing fetal loss (seven; 29%), predominantly in the first trimester (five). Moreover, fetal complications arose in two pregnancies (8%), including one pre-term birth and one with intrauterine growth retardation (IUGR) and pre-term birth. Maternal complications occurred in five pregnancies (21%), which included post-partum haemorrhage (three), pre-eclampsia (one) and gestational diabetes mellitus (one). Table I outlines additional details regarding pregnancy outcomes and disease features including treatments received both at initial diagnosis of MF and time of pregnancy.

Analysis of risk factors for fetal loss identified pre-PMF phenotype [fetal loss in seven of 16 (44%) vs. 0% in PMF/post-ET MF; P = 0·03], presence of JAK2 versus CALR mutation [four of seven (57%) vs. three of 16 (19%); P = 0·07], prior thrombosis [four of five (80%) vs. three of 19 (16%) without thrombosis; P = 0·007], and history of prior fetal loss (three of five (60%) vs. four of 19 (21%) without prior fetal loss; P = 0·10], as significant predictors of fetal loss. On the other hand, aspirin use during pregnancy was found to be protective in terms of fetal loss with only two of 14 (14%) losses as opposed to five of 10 (50%) in the absence of active therapy with aspirin (P = 0·05). Multivariable logistic regression analysis confirmed prior thrombosis history (P < 0·01) and prior fetal loss (P = 0·03) as independent risk factors, and aspirin use as being protective (P = 0·07). Factors that did not influence fetal outcome included age at pregnancy (P = 0·21), leucocyte count (P = 0·98), platelet count (P = 0·63), palpable spleen (P = 0·94) and cytoreductive therapy during pregnancy (P = 0·23). Analysis restricted to first trimester fetal losses confirmed the significant/near-significant influence of prior thrombosis history (60% vs. 11%; P = 0·03), pre-PMF phenotype (31% vs. 0%; P = 0·09) and aspirin use during pregnancy (8% vs. 36% P = 0·07). In terms of maternal complications, the only predictor was presence of JAK2 mutation as opposed to CALR mutation, with complication rates of 43% versus 13% respectively (P = 0·11). Maternal age (P = 0·25), splenomegaly (P = 0·93), prior thrombosis (P = 0·26), haemorrhage (P = 0·33), leucocyte count (P = 0·16), platelet count (P = 0·32), or treatment with aspirin (P = 0·93), systemic anticoagulants (P = 0·26) or cytoreductive agents (P = 0·32) did not appear to impact maternal complication rate.

In the present report, we share our experience with 24 pregnancies among 16 women with MF, of which 11 with pre-PMF. We were encouraged with the relatively low miscarriage rate of 29%, which is in range with what has been observed in the past with ET (30% fetal loss rate across four of the largest studies);² however, the maternal complication rate of 21% was higher in the present report, compared to what has been reported in ET (<10%).² Also, as was the case with ET,^{2, 3, 10, 11} history of prior fetal loss was predictive of subsequent loss, while use of aspirin therapy during pregnancy or presence of CALR mutations were associated with lower risk of miscarriage. A prior vascular event was detrimental to both risk of fetal loss and maternal complications, requiring special attention regarding management. Accordingly, we recommend a risk-adapted management approach based on prior vascular events and pregnancy complications.¹² Because of the relatively small number of cases in the present report, our observations should not be taken as being definitive but instead thought-provoking and in need of validation from larger studies.

Author contributions

Naseema Gangat, Paola Guglielmelli, Alessandro M. Vannucchi and Ayalew Tefferi designed the study. Paola Guglielmelli, Aref Al-Kali, Alexandra P. Wolanskyj-Spinner, John Camoriano, Mrinal M. Patnaik, Animesh Pardanani and Alessandro M. Vannucchi contributed patients. Curtis A. Hanson reviewed bone marrow pathology. Naseema Gangat and Ayalew Tefferi performed analyses and wrote the paper.

Conflict of interest

None.