The incidence of central nervous system (CNS) disease in adults with acute lymphoblastic leukemia (ALL) is 5%–15% at diagnosis and 5% at relapse with intrathecal chemotherapy (ITC).^{1, 2} CNS is a known sanctuary site of ALL, and leukemic infiltration into the CNS can lead to significant morbidity and mortality. Disease features such as hyperleukocytosis at presentation,³ common in Philadelphia chromosome-positive ALL (Ph + ALL), increase the risk of CNS relapse. The addition of BCR::ABL1 tyrosine kinase inhibitors (TKIs) to the backbone of ALL therapies, including ITC, has significantly improved survival. However, with survival prolongation, CNS relapses occurred in 8%–17% of Ph + ALL patients.^{2, 4, 5} To reduce this incidence and prevent CNS relapses, treatment protocols with hyper-fractionated cyclophosphamide, dexamethasone, vincristine, and doxorubicin alternating with high-dose methotrexate and cytarabine (HCVAD) plus a BCR::ABL1 TKI for adults with Ph + ALL were amended in 2012 to increase the number of prophylactic ITCs empirically from 8 to 12. This study assessed the rate of CNS relapses in adults with Ph + ALL treated with 8 versus 12 ITC.

We evaluated 150 consecutive patients ≥18 years of age (median: 51 years, range: 19–84), with newly diagnosed Ph + ALL treated with HCVAD ± rituximab plus imatinib, dasatinib, or ponatinib between July 2001 and January 2019. Patients with CNS disease at diagnosis were excluded. The ITC consisted mainly of methotrexate 12 mg alternating with cytarabine 100 mg given via lumbar puncture during each cycle for up to 6 cycles. Poor risk cytogenetics were defined as the presence of +der(22)t(9;22) or − 9/9p in the absence of high hyperdiploidy.⁶ Complete molecular response (CMR) was defined as the absence of quantifiable BCR::ABL1 transcripts by reverse-transcriptase quantitative polymerase chain reaction (RT-qPCR).⁷ Assessment of measurable residual disease (MRD) at complete remission (CR) was performed using multiparametric flow cytometry (FCM) and RT-qPCR for BCR::ABL1 transcripts. CNS relapse was defined by the detection of blasts or atypical cells in the cerebrospinal fluid (CSF) in at least two successive evaluations or by findings of leptomeningeal disease on imaging. A landmark analysis was performed at six months, which is the usual time to completion of ITC. Patients who died or were lost to follow-up within six months of therapy were excluded. The CNS relapse-free survival (RFS) was defined from the start of therapy until the time of CNS relapse. Patients who died or relapsed in the bone marrow were censored at time of death and systemic relapse, respectively. Survival was assessed with and without censoring for allogeneic stem cell transplantation (allo-SCT). The time to allo-SCT was handled as a time-dependent variable. The Kaplan–Meier method⁸ was used to analyze the CNS RFS. The univariate and multivariate Cox proportional hazards regressions were used to identify prognostic factors associated with CNS RFS.

Overall, 150 consecutive patients were evaluated; 106 patients (71%) received ≤8 ITC and 44 patients (29%) received >8 ITC. The patient characteristics and univariate and multivariate analysis (MVA) are summarized in the Table S1. The characteristics were similar between groups except more patients in the >8 ITC group received rituximab (43% vs. 22%, p = 0.01) and ponatinib (66% vs. 21%, p ≤ 0.001). Response outcomes to the frontline therapy in patients with ≤8 ITC and >8 ITC were similar (CR 100% versus 100%) but patients with >8 ITC achieved a higher rate of three-month CMR (70% vs. 53% p = 0.08).

The median follow-up time of the population was 107.9 months (95% CI 97.3.7–118.7), 125.8 months (95% CI 112.8–138.9) for patients who received ≤8 ITC, and 76.9 months (95% CI 67.4–86.4) for patients who received >8 ITC. CNS relapse occurred in 11 patients (7%, 4 received imatinib and 7 received dasatinib); all had received ≤8 ITC. The incidence of CNS relapse was 10% with ≤8 ITC versus 0% with >8 ITC (p = 0.023). The three- and six-year CNS RFS rates were 92% and 91% in ≤8 ITC cohort and 100% in >8 ITC cohort (three-year CNS RFS p = 0.06; six-year CNS RFS p = 0.04 (Figure 1A). These outcomes remained statistically significant after censoring for allo-SCT (p = 0.046) (Figure 1B). After adjusting for the follow-up time, the MVA showed that >8 prophylactic ITC was significantly associated with decreased rate of CNS relapse (p = 0.03; HR = 0.104, 95% CI: 0.001–0.801).

CNS relapses occurred after a median of 17 months (range 6–94). In four of 11 (36%) patients, CNS relapse co-occurred with systemic relapse. One patient experienced systemic relapse after isolated CNS relapse. At time of CNS relapse, BCR::ABL1 transcripts were detected in the peripheral blood or bone marrow of all patients. Treatment of CNS relapse varied from systemic chemotherapy to craniospinal irradiation, but all continued TKI with ITC, consisting of cytarabine, alternating with methotrexate, with hydrocortisone, or combination given twice weekly until CNS clearance. All patients achieved CNS clearance with 4 (36%) experiencing subsequent CNS relapses. The median survival from time of first CNS relapse was 14.7 months (95% CI 8.6–20.8). At the time of data cutoff, only one patient remained alive after CNS relapse. Five patients (50%) died due to infectious complications, one (10%) of intracranial hemorrhage, two (20%) of cardiac arrest, and two from progressive leukemia. Five of the 11 (45%) patients underwent allo-SCT during their treatment course with one allo-SCT preceding CNS relapse. The characteristics and dynamics of CNS relapse for individual patients are shown in Figure S2.

This study demonstrates that the incorporation of 12 prophylactic ITC to HCVAD in adults with Ph + ALL significantly reduces the risk of CNS relapse. No CNS relapses occurred with ponatinib, possibly due to its ability to induce early CMR and eradicate the malignant clone systemically rather than its debatable activity in the CNS. HCVAD plus ponatinib has demonstrated superiority over dasatinib⁹ and imatinib in achieving a CMR at three months,¹⁰ a prognostic marker for improved outcomes.^{7, 11} Penetration of all three TKIs across the blood–brain barrier (BBB) have been studied.^12-15 Ponatinib's potential to cross the BBB has been demonstrated in mouse models and scant case-reports exist eluding to its activity. However, in-human studies are necessary to establish its optimal concentration and exposure needed for a meaningful clinical benefit in the CNS.¹³ On the other hand, dasatinib has been well described to have good CNS activity with its efficacy proportional to dosing.^{12, 14} Therefore, we compared dasatinib to imatinib and ponatinib which was not significantly associated with less CNS relapses by MVA (p = 0.225; HR: 2.04, 95% CI: 0.46–7.190). We observed CNS relapses in both imatinib and dasatinib groups, specifically when ≤8 ITCs were administered. The number of ITCs administered (>8 ITCs, median of 12 ITCs) was the most consistent variable among the ponatinib group as 24 of 51 patients (47%) switched to dasatinib (n = 19; 79%) or imatinib (n = 5; 21%) after a median of 27 months due to toxicity, intolerance, or its withdrawal from the market in 2013, which is reflective of real-world practice. Until the impact of ponatinib and its interaction with ITC can be investigated further, incorporating 12 ITC should be considered to reduce the risk of CNS relapse in adults Ph + ALL, regardless of the TKI used.

Recent advances in chemotherapy-free regimens, such as blinatumomab plus dasatinib¹⁶ or ponatinib¹⁷ incorporating 12 ITC, have shown promising results; however, the dasatinib-blinatumomab study (N = 63) reports four CNS relapses (6%), at a median follow-up of 27 months. Although total ITCs administered was not reported, its plausible that patients received <12 ITC since majority of the patients received an allograft shortly after the first complete hematologic response. A longer follow-up is needed to determine whether the risk of CNS relapse increases in the absence of CNS-active systemic chemotherapy (such as high-dose methotrexate and high-dose cytarabine), whether >12 ITC are needed, and whether the total number of ITC doses should be influenced by the TKI received.

In conclusion, ITC is essential in preventing CNS disease in Ph + ALL, and effectiveness depends on the total number of ITC administered. Our study shows that 12 ITCs should be administered with HCVAD plus a BCR::ABL1 TKI to reduce the risk of CNS relapse.

成人急性淋巴细胞白血病 (ALL) 中枢神经系统 (CNS) 疾病的发病率在诊断时为 5%–15%，在鞘内化疗 (ITC) 复发时为 5%。^{1, 2} CNS 是已知的 ALL 避难所，白血病浸润到 CNS 可导致显着的发病率和死亡率。就诊时白细胞增多症等疾病特征，³在费城染色体阳性 ALL (Ph + ALL) 中很常见，会增加 CNS 复发的风险。将 BCR::ABL1 酪氨酸激酶抑制剂 (TKI) 添加到 ALL 疗法（包括 ITC）的主干中，可显着提高生存率。然而，随着生存期的延长，8%–17% 的 Ph + ALL 患者出现 CNS 复发。^{2, 4, 5}为了降低这种发生率并防止 CNS 复发，对 Ph + ALL 成人患者使用超分割环磷酰胺、地塞米松、长春新碱和多柔比星与高剂量甲氨蝶呤和阿糖胞苷 (HCVAD) 以及 BCR::ABL1 TKI 交替治疗方案进行了修订2012 年根据经验将预防性 ITC 的数量从 8 个增加到 12 个。这项研究评估了 Ph + ALL 成人中 8 对 12 ITC 治疗的中枢神经系统复发率。

我们评估了 150 名年龄≥18 岁（中位数：51 岁，范围：19-84 岁）的连续患者，这些患者在 2001 年 7 月至 2019 年 1 月期间接受了 HCVAD ± 利妥昔单抗加伊马替尼、达沙替尼或普纳替尼治疗的新诊断 Ph + ALL。诊断时的 CNS 疾病被排除在外。ITC 主要包括甲氨蝶呤 12 mg 与阿糖胞苷 100 mg 交替给药，每个周期通过腰椎穿刺给药，最多 6 个周期。不良风险细胞遗传学被定义为在没有高超二倍体的情况下存在 +der(22)t(9;22) 或 - 9/9p。⁶完全分子反应 (CMR) 被定义为通过逆转录酶定量聚合酶链反应 (RT-qPCR)没有可量化的BCR::ABL1转录本。⁷使用多参数流式细胞术 (FCM) 和 RT-qPCR 评估完全缓解 (CR) 时的可测量残留病灶 (MRD) BCR::ABL1成绩单。CNS 复发的定义是在至少两次连续评估中检测到脑脊液 (CSF) 中的原始细胞或非典型细胞，或通过影像学发现软脑膜疾病。里程碑分析在六个月时进行，这是 ITC 完成的通常时间。治疗后六个月内死亡或失访的患者被排除在外。CNS 无复发生存期 (RFS) 定义为从治疗开始到 CNS 复发时间。在骨髓中死亡或复发的患者分别在死亡和全身复发时进行审查。在对同种异体干细胞移植 (allo-SCT) 进行审查和不审查的情况下评估生存率。allo-SCT 的时间作为时间相关变量处理。Kaplan–Meier 方法⁸用于分析 CNS RFS。单变量和多变量 Cox 比例风险回归用于确定与 CNS RFS 相关的预后因素。

总体而言，对 150 名连续患者进行了评估；106 名患者 (71%) 接受了 ≤8 个 ITC，44 名患者 (29%) 接受了 >8 个 ITC。表 S1 总结了患者特征以及单变量和多变量分析 (MVA)。除了 >8 ITC 组中更多患者接受利妥昔单抗（43% 对 22%，p  = 0.01）和普纳替尼（66% 对 21%，p  ≤ 0.001）外，各组间的特征相似。ITC ≤8 和 ITC >8 的患者对一线治疗的反应结果相似（CR 100% 对 100%），但 ITC >8 的患者达到更高的三个月 CMR 率（70% 对 53% p  = 0.08）。

人群的中位随访时间为 107.9 个月（95% CI 97.3.7–118.7），接受 ≤8 ITC 的患者为 125.8 个月（95% CI 112.8–138.9），76.9 个月（95% CI 67.4– 86.4) 对于接受 >8 ITC 的患者。11 名患者发生 CNS 复发（7%，4 名接受伊马替尼，7 名接受达沙替尼）；所有人都收到了≤8个ITC。≤8 ITC 的 CNS 复发发生率为 10%，大于 8 ITC 的发生率为 0% ( p  = 0.023)。≤8 ITC 队列的三年和六年 CNS RFS 率为 92% 和 91%，>8 ITC 队列为 100%（三年 CNS RFS p  = 0.06；六年 CNS RFS p  = 0.04（图 1A） ). 这些结果在审查异基因 SCT 后仍然具有统计学意义 ( p = 0.046) (图 1B)。调整随访时间后，MVA 显示 >8 的预防性 ITC 与 CNS 复发率降低显着相关（p  = 0.03；HR = 0.104，95% CI：0.001–0.801）。

CNS 复发发生在中位 17 个月后（范围 6-94）。在 11 名患者中的 4 名 (36%) 中，CNS 复发与全身复发同时发生。一名患者在孤立的 CNS 复发后出现全身性复发。在 CNS 复发时，BCR::ABL1在所有患者的外周血或骨髓中检测到转录本。CNS 复发的治疗从全身化疗到颅脊髓放疗不等，但所有患者均继续使用 ITC 进行 TKI，包括阿糖胞苷，与甲氨蝶呤、氢化可的松交替使用，或每周两次联合用药，直至 CNS 清除。所有患者均获得 CNS 清除，其中 4 名 (36%) 患者随后出现 CNS 复发。从第一次 CNS 复发开始的中位生存期为 14.7 个月 (95% CI 8.6–20.8)。在数据截止时，只有一名患者在 CNS 复发后还活着。五名患者 (50%) 死于感染性并发症，一名 (10%) 死于颅内出血，两名 (20%) 死于心脏骤停，两名死于进行性白血病。11 名患者中有 5 名 (45%) 在治疗过程中接受了异基因 SCT，其中 1 名患者在 CNS 复发前接受了异基因 SCT。个别患者 CNS 复发的特征和动态如图 S2 所示。

这项研究表明，在患有 Ph + ALL 的成人中，将 12 种预防性 ITC 与 HCVAD 结合可显着降低 CNS 复发的风险。普纳替尼没有发生 CNS 复发，这可能是因为它能够诱导早期 CMR 并全身性地根除恶性克隆，而不是它在 CNS 中有争议的活性。HCVAD 加普纳替尼已证明在三个月时达到 CMR ¹⁰方面优于达沙替尼⁹和伊马替尼，CMR是改善结果的预后标志。^{7, 11}已经研究了所有三种 TKI 穿过血脑屏障 (BBB) 的渗透性。^12-15Ponatinib 跨越 BBB 的潜力已在小鼠模型中得到证实，并且存在逃避其活动的少量病例报告。然而，有必要进行人体研究以确定其在 CNS 中获得有意义的临床益处所需的最佳浓度和暴露量。¹³另一方面，达沙替尼已被充分描述为具有良好的中枢神经系统活性，其疗效与剂量成正比。^{12, 14}因此，我们将达沙替尼与伊马替尼和普纳替尼进行了比较，后者与 MVA 引起的中枢神经系统复发较少没有显着相关性 ( p = 0.225；HR：2.04，95% CI：0.46–7.190）。我们在伊马替尼和达沙替尼组均观察到 CNS 复发，特别是当施用 ≤ 8 个 ITC 时。给予的 ITC 数量（>8 个 ITC，中位数为 12 个 ITC）是普纳替尼组中最一致的变量，因为 51 名患者中有 24 名 (47%) 转用达沙替尼（n  = 19；79%）或伊马替尼（n  = 5 ; 21%) 在中位数 27 个月后由于毒性、不耐受或 2013 年退出市场，这反映了现实世界的实践。在进一步研究 ponatinib 的影响及其与 ITC 的相互作用之前，无论使用何种 TKI，都应考虑加入 12 ITC 以降低成人 Ph + ALL 中枢神经系统复发的风险。

无化疗方案的最新进展，例如博纳吐单抗加达沙替尼¹⁶或含有 12 ITC 的普纳替尼¹⁷，已显示出可喜的结果；然而，达沙替尼-blinatumomab 研究（N = 63) 报告四次 CNS 复发 (6%)，中位随访时间为 27 个月。尽管未报告总的 ITC 施用量，但患者接受的 ITC 数量少于 12 个是有道理的，因为大多数患者在第一次完全血液学反应后不久就接受了同种异体移植。需要更长时间的随访来确定在没有 CNS 活性全身化疗（如大剂量甲氨蝶呤和大剂量阿糖胞苷）的情况下 CNS 复发的风险是否增加，是否需要 >12 ITC，以及总ITC 剂量的数量应该受到接受的 TKI 的影响。

总之，ITC 对于预防 Ph + ALL 中的 CNS 疾病至关重要，有效性取决于施用的 ITC 总数。我们的研究表明，12 个 ITC 应与 HCVAD 加 BCR::ABL1 TKI 一起使用，以降低 CNS 复发的风险。