Despite all the evolution of chronic myeloid leukemia (CML) treatment, disease management during pregnancy remains a significant challenge, and the knowledge about the impact of pregnancy on long-term outcomes of CML is scarce. Recently, criteria for the selection of patients for treatment-free remission (TFR) were established in international guidelines, encouraging women to get pregnant in a planned way. Thus, not all pregnancies are planned and just half of the patients in TFR keep the major molecular response (MMR) for the long term. This study aims to evaluate the impact of pregnancy on CML outcomes in the female population of childbearing age. CML management during pregnancy and maternal-fetal complications is discussed based on this series of cases.

From January 1991 to September 2018, all women under 45 years old and CML diagnosis admitted at Hospital Complex of the University of São Paulo, Brazil, were included in this retrospective study, approved by the local Research Ethics Committee.

The patients were split into control group (C), which included women who did not become pregnant after diagnosing CML and another group with pregnant women (P). Age, disease stage, and Sokal score at diagnosis; the first line of tyrosine kinase inhibitor (TKI), response and compliance to treatment, and switch to a second line TKI during the follow-up were assessed. Poor compliance was defined as abandonment of treatment for more than 30 days, nonattendance to the planned medical appointment or rise in the BCR/ABL transcripts number (> 1 log) with recovery without changing the TKI. Pregnancies and their complications, and fetal malformations were events of interest. Planned pregnancy was considered the report of the desire to get pregnant with TKI interruption before conception. For group P, the following data were also collected: TKI use and CML status at conception, CML treatment, and response during pregnancy and after delivery. Accelerated phase (AP), blast crisis (BC), or death and the disease status at last attendance were outcomes of interest.

Details of the statistical analysis are available in the online supplement.

One hundred patients were assessed for eligibility and two were excluded (allogeneic transplant and lose the follow up before TKI therapy), 63 were included in group C and 35 in group P. Nine patients had two or more pregnancies, totalizing 46, and only 6 pregnancies (in 4 women) were planned. There were no significant differences between groups considering age, CML phase, Sokal score, first-line TKI, and TKI switching (Table S1).

Poor compliance was reported in 28% of all cohort, 22% (n = 14) of group C, and 40% in group P (n = 14) (p = .052). Furthermore, it was higher in women with unplanned pregnancies (52%) than in control group (p = .009).

The median follow-up was 11.7 years (range 2–29). At 10 years, progression-free survival (PFS) was 93% (CI 95%: 86.5%–99.9%) and 88% (CI 95%: 77.5%–99.8%) and overall survival (OS) was 96.5% (CI 95%: 91.8%–100%) and 90.6% (CI 95%: 81.1%–100%), in group C and P, respectively, with no statistically significant differences. PFS and OS were lower in the subgroup of unplanned pregnancy (UnP) with poor compliance compared to group C (Figure 1). PFS was also lower in women who got pregnant after CML diagnosis and conceived while not in MMR (70.0%, 95% CI, 46.7%–100%) (Figure S1). Just poor compliance was identified as an independent risk factor (HR: 11.7, 95% CI: 2.6–52.0, p = .001 for PFS).

The proportion of women achieving sustained MMR was similar between the groups. In group P, the rate of evolution to AP and BC was higher (14.2% versus 6%). Twelve deaths were reported: 6/63 from group C (9.5%) and 6/35 from group P (17%) (p = .3), having disease progression as the leading cause (75%), mostly related to poor adherence.

Fifteen women had the diagnosis of CML during pregnancy and nine of them received CML treatment with Interferon-alpha (IFN) and only one required leukapheresis. Imatinib was started after delivery and 6/15 (40%) were in MMR at 12 months. At the last assessment, 9/15 women (60%) were in deep MR (dMR) and 4/15 were in MMR (27%).

Thirty-one pregnancies have occurred after the CML diagnosis. Sixteen were in MMR at conception (only seven were planned) and half of them received IFN to preserve MMR, however, loss of MMR has occurred in 7/8. During pregnancy, CML was monitored using the quantitative polymerase chain reaction (PCR) for BCR-ABL, collected every 30–60 days, until the loss of MMR or delivery. Despite a high rate of loss of molecular response (10/16), including four loss of hematological response (HR), 9/10 recovered MMR in 18 months after TKI resumption.

Among 12 women with no MMR at the conception time, four were also without HR. CML treatment were: IFN (10), leukapheresis (2), and hydroxyurea (1). 6/12 has died due to disease progression.

Seventeen of all pregnancies (37%) ended with complications, including nine miscarriages, three premature births, one fetal distress, three hypertensive diseases specific to pregnancy, and one placental abruption. Nineteen pregnancies were conceived while taking TKI: imatinib (9), dasatinib (7), and nilotinib (3) and 10 (52%) evolved with complications, including two fetal malformations (encephalocele with the absence of cerebellar vermis and hydrocele).

This is the first study comparing the long-term impact of pregnancy in childbearing-age women with CML. In a small cohort of 16 cases, Lasica et al. compared women with planned versus unplanned pregnancies and did not observe the worst outcome between groups, similar to our population.¹

Compliance was the only independent predictor of progression or death in our cohort, however, adherence measurement is a challenge in clinical practice and clinical trials worldwide. Although the overall poor compliance rate in our study was 28%, this data were obtained retrospectively and probably is underestimated. Among pregnant women, this number increases to 40% and up to 52% in the UnP subgroup, associated with worst PFS, similar to the study by Chen et al.²

Considering the disease status at the conception, women with MMR–dMR, despite de high rate of loss of response after TKI suspension (63%), 90% recovered it after the TKI resumption, and none had disease progression, suggesting that the conception in MMR seems to be safe, despite the UnP. Likewise, the Italian GIMEMA registry reports 76.5% loss of MMR in the 60 pregnancies, but none of them showed disease progression during the follow-up.³ Our data suggest worse PFS for women who conceived without adequate CML control but is necessary for a larger cohort to analyze the real impact of CML status at conception on outcomes.

Only seven pregnancies were planned, with interruption of the TKI for conception in patients who were in MMR. With TFR studies, new options are opening up for women who wish to become pregnant. Some questions are still open in the context of women of childbearing age and desire to become pregnant, such as the choice of the first-line TKI considering the requirement for dMR for stopping treatment, better time and form of conception (natural or in vitro), better management of CML during pregnancy, among others.

There is no consensus on the best treatment of CML during pregnancy. The recommendation of immediate interruption of the TKI is universal, as long as the CML status allows and continuation of the pregnancy is possible. IFN inhibits cell proliferation but does not interfere with DNA synthesis and can be used safely after the first trimester of pregnancy. In our population, preemptive use was made for half of the women in MMR to preserve the molecular response, but 88% of them have lost it and no benefit was observed. Both imatinib and nilotinib may be considered after the first trimester.

The abortion rate is around 10% in healthy women worldwide.⁴ We did not observe a higher rate of abortion or premature birth compared to healthy women. However, 52% of women who conceived while in TKI had pregnancy complications or fetal malformation, higher than a previous literature review that found 33% of maternal/fetal complications.⁵ This difference may take into account the use of second-generation TKI in half of our population.

The incidence of congenital anomalies in pregnancies in the general population is estimated at 3%–5%, compared to 10.5% in our cohort, which may be related to the teratogenic effect of TKIs, considering that 45% of pregnancies reported are unplanned, conceived while in TKI use.⁶

Limitations of our study are the retrospective design, which results in underestimating the real adherence measurement, loose important information for Sokal score and social demographic analysis; and the small number of events. Despite the difficulties in carrying out multicenter prospective studies in our country for several reasons, mainly due to the absence of a national register of CML, they are required to improve information about outcomes and management of CML in pregnancy. Efforts must be done and local protocol for childbearing age women should be encouraged.

In conclusion, pregnancy itself does not seem to interfere in the outcomes of CML in the long term. Although treatment adherence is still the main challenge, this study has shown that UnP is a warning sign of poor compliance, and these women deserve special attention.

尽管慢性粒细胞白血病 (CML) 治疗取得了所有进展，但妊娠期间的疾病管理仍然是一项重大挑战，而且关于妊娠对 CML 长期结果影响的知识很少。近期，国际指南制定了免治疗缓解（TFR）患者的选择标准，鼓励女性有计划地怀孕。因此，并非所有怀孕都是计划好的，只有一半的 TFR 患者长期保持主要分子反应 (MMR)。本研究旨在评估妊娠对育龄女性人群 CML 结局的影响。基于这一系列病例讨论妊娠期CML管理和母胎并发症。

从 1991 年 1 月到 2018 年 9 月，所有在巴西圣保罗大学医院综合院收治的 45 岁以下且诊断为 CML 的女性都被纳入了这项经当地研究伦理委员会批准的回顾性研究。

将患者分为对照组（C），其中包括诊断出CML后未怀孕的女性和另一组孕妇（P）。诊断时的年龄、疾病阶段和 Sokal 评分；评估了一线酪氨酸激酶抑制剂 (TKI)、对治疗的反应和依从性，以及在随访期间改用二线 TKI。依从性差被定义为放弃治疗超过 30 天、未参加计划的医疗预约或 BCR/ABL 成绩单数量增加（> 1 log）随着恢复而不改变 TKI。妊娠及其并发症和胎儿畸形是令人感兴趣的事件。计划怀孕被认为是在受孕前中断 TKI 怀孕的愿望的报告。对于 P 组，还收集了以下数据：TKI 的使用和受孕时的 CML 状态、CML 治疗以及怀孕期间和分娩后的反应。加速期 (AP)、爆炸危机 (BC) 或死亡以及最后一次就诊时的疾病状态是令人感兴趣的结果。

在线补充资料中提供了统计分析的详细信息。

评估了 100 名患者的资格，排除了 2 名（同种异体移植和 TKI 治疗前失访），C 组 63 名，P 组 35 名。9 名患者有 2 次或以上妊娠，总共 46 例，只有 6 例计划怀孕（4 名妇女）。考虑到年龄、CML 阶段、Sokal 评分、一线 TKI 和 TKI 转换，各组之间没有显着差异（表 S1）。

据报道，所有队列中有 28%、C 组 22%（n  = 14）和 P 组 40%（n  = 14）（p  = .052）的依从性差。此外，意外怀孕的女性（52%）高于对照组（p  = .009）。

中位随访时间为 11.7 年（范围 2-29）。10 年时，无进展生存期 (PFS) 为 93% (CI 95%: 86.5%–99.9%) 和 88% (CI 95%: 77.5%–99.8%)，总生存期 (OS) 为 96.5% (CI C组和P组分别为95%：91.8%–100%）和90.6%（CI 95%：81.1%–100%），差异无统计学意义。与 C 组相比，依从性差的非计划妊娠 (UnP) 亚组的 PFS 和 OS 较低（图 1）。在 CML 诊断后怀孕且未在 MMR 中受孕的女性的 PFS 也较低（70.0%，95% CI，46.7%–100%）（图 S1）。仅差的依从性被确定为一个独立的风险因素（HR：11.7，95% CI：2.6-52.0， PFS 的p  = .001）。

各组之间实现持续 MMR 的女性比例相似。在 P 组中，进化为 AP 和 BC 的速率更高（14.2% 对 6%）。报告了 12 例死亡：C 组 6/63（9.5%）和 P 组 6/35（17%）（p  = .3），疾病进展是主要原因（75%），主要与依从性差有关.

15 名女性在怀孕期间被诊断为 CML，其中 9 名接受了干扰素-α (IFN) 治疗 CML，只有 1 名需要进行白细胞分离术。伊马替尼在分娩后开始使用，6/15 (40%) 在 12 个月时处于 MMR。在最后一次评估中，9/15 女性 (60%) 处于深部 MR (dMR) 中，4/15 处于 MMR (27%) 中。

在 CML 诊断后发生了 31 次怀孕。16 人在受孕时处于 MMR 中（仅计划了 7 人），其中一半接受了 IFN 以保持 MMR，然而，7/8 发生了 MMR 的丧失。在怀孕期间，使用 BCR-ABL 的定量聚合酶链反应 (PCR) 监测 CML，每 30-60 天收集一次，直到 MMR 消失或分娩。尽管分子反应丧失率很高 (10/16)，包括四次血液学反应 (HR) 丧失，9/10 在 TKI 恢复后的 18 个月内恢复了 MMR。

在受孕时没有 MMR 的 12 名女性中，有 4 名也没有 HR。CML 治疗是：IFN (10)、白细胞分离术 (2) 和羟基脲 (1)。6/12 因疾病进展而死亡。

所有妊娠中有 17 例（37%）以并发症告终，包括 9 例流产、3 例早产、1 例胎儿窘迫、3 种妊娠特有的高血压疾病和 1 例胎盘早剥。服用 TKI 时怀孕了 19 例：伊马替尼 (9)、达沙替尼 (7) 和尼罗替尼 (3)，10 例 (52%) 出现并发症，包括两个胎儿畸形（无小脑蚓部和鞘膜积液的脑膨出）。

这是第一项比较妊娠对患有 CML 的育龄妇女的长期影响的研究。在一个 16 例病例的小队列中，Lasica 等人。比较了计划怀孕和计划外怀孕的女性，并且没有观察到组间最差的结果，类似于我们的人群。¹

在我们的队列中，依从性是进展或死亡的唯一独立预测因子，然而，依从性测量是全球临床实践和临床试验中的一项挑战。尽管我们研究中的总体依从性差为 28%，但该数据是回顾性获得的，可能被低估了。在孕妇中，这一数字增加到 40%，在 UnP 亚组中增加到 52%，与最差的 PFS 相关，类似于 Chen 等人的研究。²

考虑到受孕时的疾病状态，MMR-dMR 的女性尽管在 TKI 暂停后失去反应的比例很高（63%），但在 TKI 恢复后恢复的比例高达 90%，并且没有人出现疾病进展，这表明在尽管有 UnP，但 MMR 似乎是安全的。同样，意大利 GIMEMA 登记处报告 60 例妊娠中 MMR 损失 76.5%，但在随访期间均未显示疾病进展。³我们的数据表明，在没有充分控制 CML 的情况下受孕的女性 PFS 较差，但对于更大的队列分析受孕时 CML 状态对结局的真正影响是必要的。

仅计划了 7 次怀孕，在 MMR 患者中中断了 TKI 受孕。通过 TFR 研究，为希望怀孕的女性提供了新的选择。在育龄妇女和希望怀孕的情况下，一些问题仍然悬而未决，例如一线 TKI 的选择考虑到 dMR 停止治疗的要求、更好的时间和受孕形式（自然或体外） ，在怀孕期间更好地管理 CML，等等。

关于妊娠期 CML 的最佳治疗尚无共识。只要 CML 状态允许并且可以继续妊娠，立即中断 TKI 的建议是普遍的。干扰素抑制细胞增殖，但不干扰 DNA 合成，可在妊娠头三个月后安全使用。在我们的人群中，一半的女性在 MMR 中抢先使用以保持分子反应，但其中 88% 的女性已经失去了分子反应，并且没有观察到任何益处。伊马替尼和尼罗替尼都可以在妊娠头三个月后考虑。

全世界健康女性的堕胎率约为 10%。⁴与健康女性相比，我们没有观察到更高的流产或早产率。然而，在 TKI 期间受孕的女性中有 52% 有妊娠并发症或胎儿畸形，高于之前发现的 33% 的母胎并发症的文献回顾。⁵这种差异可能考虑到我们一半人口使用第二代 TKI。

一般人群中先天性异常的发生率估计为 3%–5%，而我们的队列中为 10.5%，这可能与 TKI 的致畸作用有关，因为报告的妊娠中有 45% 是计划外的、受孕的在 TKI 使用时。⁶

我们研究的局限性在于回顾性设计，这导致低估了真正的依从性测量，松散了索卡尔评分和社会人口统计分析的重要信息；和少量的事件。尽管由于几个原因在我国开展多中心前瞻性研究存在困难，主要是由于缺乏国家 CML 登记册，但仍需要改进有关妊娠期 CML 结局和管理的信息。必须作出努力，并应鼓励育龄妇女的当地协议。

总之，从长远来看，怀孕本身似乎不会干扰 CML 的结果。尽管治疗依从性仍然是主要挑战，但这项研究表明，UnP 是依从性差的警告信号，这些女性值得特别关注。