Three anti-CD19 chimeric antigen receptor (CAR) T-cell therapy products, have been approved by the US Food and Drug Administration (FDA) since 2017 for patients with large B-cell lymphoma (LBCL) relapsing or refractory (R/R) to two more lines of therapy.^1-3 High-dose chemotherapy and autologous stem cell transplant (ASCT) remains the standard of care for patients with R/R LBCL who demostrate chemosensitivity (complete response (CR) or partial response (PR)) to second-line chemotherapy (chemo). Ongoing clinical trials are currently comparing ASCT to CAR-T in the second-line setting. Patients treated on the registration trials and in the real-world setting were heavily pre-treated (≥ 3 prior lines of therapy: ZUMA-1 – 69%, JULIET- 51%, US CAR-T consortium- 74%).^{1, 2, 4} Due to the aggressive nature of this heavily pre-treated R/R population, the timing of CAR-T referral is important. The median time from apheresis to CAR-T cell infusion in one real-world experience was 28–44 days depending on the product.⁵ Other factors contributing to overall wait time include the availability of specialists and treatment centers, travel, especially for those in remote areas, or international patients with lack of access to CAR-T, and insurance approval. While waiting, patients may require additional treatment to maintain control over the disease. Increasing wait times of receiving CAR-T therapy doubled the predicted 1-year mortality rate in a study using a health system-level discrete simulation model.⁶ The objective of our study was to identify how the number of prior lines of therapy impacted ASCT and CAR-T patients' outcomes.

A retrospective review of patients who received ASCT and/or axicabtagene ciloleucel (axi-cel) for R/R aggressive LBCL from June 2016 – April 2020 at Mayo Clinic, Rochester, was performed. The study was conducted in accordance with the Declaration of Helsinki and was approved by Mayo Clinic IRB. Patients were managed, including the decision for bridging therapy and the timing of imaging pre-ASCT and post-ASCT or CAR-T based on the institutional guidelines and treating physician's discretion. Additional details on the methodology regarding assessment of baseline characteristics, definitions of lines of therapy, and response assessment are provided in supplement. Different comparisons made between groups were based on the lines of therapy before ASCT or CAR-T encountered in the real-world setting (Figure S1, consort diagram). Patients underwent ASCT either after two (ASCT [2]) or three (ASCT [3]) lines of chemo. As for CAR-T groups, the earliest possible indication patients could undergo CAR-T (CAR-T[early]) was either after two lines of chemotherapy (CAR-T [2]) and/or after ASCT [2]. Those not meeting the CAR-T[early] criteria were grouped as CAR-T[late]. Additionally, patients could undergo CAR-T immediately post ASCT relapse (CAR-T[ASCT]) without any interval treatment. This group comprised patients from both ASCT [2] and ASCT [3] groups. Those not meeting the CAR-T[ASCT] criteria were grouped as CAR-T[other], as these required additional therapy between ASCT and CAR-T. All the analyses were performed in R3.6.3, p values were two-sided, and the significance level was set at <0.05.

A total of 105 patients underwent ASCT, and 62 underwent CAR-T therapy in the study period. Of the ASCT patients, 87 received ASCT after two lines of chemotherapy (ASCT [2]) and 18 received ASCT after three lines of chemotherapy (ASCT [3]). The baseline characteristics of the entire ASCT and CAR-T cohorts, are shown in Table S1. The comparison of ASCT [2], and ASCT [3] groups is shown in Table S2. Twenty-six patients (25%) of the ASCT group (n = 19, ASCT [2] and n = 7, ASCT [3]) received CAR-T as subsequent therapy. All the characteristics were comparable between the two groups except the ASCT [3] group had a higher percentage of patients with progressive disease after first-line (44% vs 21%, p < 0.001) and second line therapies. With a median follow up of 19.7 months, the 1-year event-free survival (EFS) rate was significantly higher for the ASCT [2] group 67%, (95%CI 58%–78%) vs ASCT [3] 44%, (95%CI 27%–75%; p = 0.015), Figure 1(A). However, there was no significant difference noted in the 1-year overall survival (OS) rates (Figure 1(B)).

Among the CAR-T recipients (n = 62), 28 (45%) patients had undergone previous ASCT. Only 25/62 (40%) patients received CAR-T at the earliest possible indication (CAR-T[early]), of which 15 patients belonged to the CAR-T [2] group and 10 patients to the ASCT [2] group (Table 1). The baseline characteristics of the CAR-T[early] vs CAR-T[late] groups were comparable. At a median follow-up of 12.1 months in the CAR-T cohort, there was a trend towards improved 1-year EFS (48%, (95%CI 32–72) vs CAR-T [late] 30%, (95%CI 18%–49%) p = 0.055) and OS (75%, (95%CI 60–95) vs CAR-T [late] 56%, (95%CI 42%–76%) p = 0.053) in the CAR-T[early] group (Figure 1(C) and (D)). Next we compared CAR-T[ASCT] (n = 14, 10 - ASCT [2], and four - ASCT [3]) and CAR-T[other] (n = 48) groups. A significantly higher patient proportion in the CAR-T[other] group (75% vs 43%, p = 0.03) required bridging therapy before CAR-T infusion (Table 1). The CAR-T[ASCT] group had a significantly higher 1-year OS rate of 77% (95%CI 60–100) compared to 60% (95%CI 47–76) in the CAR-T[other] group (p = 0.039), Figures 1(E) and (F). Lastly, ASCT [3] group (n = 18) and CAR-T [2] group (n = 15) were analyzed and both groups were comparable (Table S3), except CAR-T [2] group had significantly more patients with progressive disease to second line chemo (87% vs 44%, p = 0.01), and a shorter median time from diagnosis to infusion. At a median follow-up of 9.9 months, there was no significant difference noted in 1-year EFS or OS rates between the two groups (Figure 1(G) and (H)).

This report provides clinical outcomes in a single large academic center for both ASCT and CAR-T based on prior lines of therapy encountered clinically in the real world. Multiple studies have attempted to identify the patient, disease, and product-related factors that could predict durable responses to CAR-T, to allow for better patient selection. We attempted to determine when CAR-T would optimally be incorporated based on prior lines of therapy encountered in the clinical setting, other than indicated per the FDA label. We found that 1-year EFS rate was longer for ASCT [2] vs. ASCT [3] patients (67% vs 44%). The CORAL trial also demonstrated a progressively shorter EFS and OS with subsequent lines of therapy. We, however, did not identify a difference in the 1-year OS rate between ASCT [2] and ASCT [3] groups, likely owing to small numbers and the use of subsequent therapies in the ASCT [3] group (7/18). In addition, the ASCT [3] group also comprised of patients with a more primary refractory disease, a group known to have poor outcomes. The CAR-T (early) group showed a trend towards a better 1-year EFS and OS and CAR-T[ASCT] group had significantly improved OS likely suggesting better disease biology as chemosensitivity is required to proceed with ASCT. This was also evident based on an increased need for bridging therapy in the CAR-T(other) group. Sermer et al. have previously compared CAR-T outcomes to alternate therapies beyond two prior treatment lines and showed worsening CAR-T outcomes with successive lines of therapy.⁷ The alternate therapies group in the study however, comprised of patients treated over a long period (2001–2017), and included treatments other than ASCT. A recent analysis by Locke et al. identified higher tumor burden, lower CAR-T cell expansion, a proinflammatory tumor microenvironment, and increasing doubling time of CAR-T cells as factors associated with a lack of durable response. These factors were notably pronounced with an increasing number of lines of therapy and supported the observation that CAR-T efficacy decreases with increasing lines of therapy.⁸ Worse outcomes with successive lines of treatment are likely from a combination of increased disease refractoriness, high-risk disease features, and worse CAR-T cell product fitness. The degree that each of these and other factors contribute to worsening outcomes is yet to be determined. Our cohort had only 40% of patients (25/62) referred at the earliest possible indication for CAR-T per the FDA label. Early referral, therefore, is key to improving CAR-T outcomes to avoid logistic delays and need for bridging therapy. For patients achieving less than a CR after first salvage, whether to attempt another salvage or proceed with CAR-T remains unanswered. Large registry studies are currently ongoing in this direction. Our analysis comparing ASCT(3) to CAR-T(2) did not identify a difference in the 1-year EFS or OS rates. This is possibly due to a higher proportion of chemo-refractory disease to second-line treatment in the CAR-T(2) group, shorter time from diagnosis to infusion and overall small patient numbers in the cohort. This study is limited by its retrospective nature, small patient numbers and short median follow-up time which likely have some bearing on the lack of survival differences in the comparisons made. However, comparing ASCT and CAR-T cohorts over the same time period avoids the effect of differences in supportive care and a single-center analysis provides a more homogenous cohort with respect to practice patterns. Larger cohorts with longer follow-up are needed to understand the implication of sequencing of these treatment modalities. Referral for earlier use of CAR-T remains low and needs to be emphasized to avoid delays and improve access to a potentially curative therapy.

三款抗CD19嵌合抗原受体（CAR）T细胞疗法产品，自2017年起获美国食品药品监督管理局（FDA）批准用于治疗复发性或难治性（R/R）大B细胞淋巴瘤（LBCL）患者到另外两条线治疗。^1-3大剂量化疗和自体干细胞移植 (ASCT) 仍然是 R/R LBCL 患者的标准护理，这些患者对二线化疗（化疗）。目前正在进行的临床试验正在二线环境中比较 ASCT 与 CAR-T。在注册试验和现实环境中接受治疗的患者都接受了大量的预处理（≥ 3 种先前的治疗线：ZUMA-1 – 69%，JULIET- 51%，美国 CAR-T 联盟- 74%）。^1、2、4由于这个经过大量预处理的 R/R 人群的攻击性，CAR-T 转诊的时机很重要。在一项真实世界的经验中，从单采到 CAR-T 细胞输注的中位时间为 28-44 天，具体取决于产品。⁵影响总体等待时间的其他因素包括专家和治疗中心的可用性、旅行，尤其是偏远地区的人，或无法获得 CAR-T 的国际患者，以及保险批准。在等待期间，患者可能需要额外的治疗以保持对疾病的控制。在一项使用卫生系统级离散模拟模型的研究中，增加接受 CAR-T 治疗的等待时间使预测的 1 年死亡率增加了一倍。⁶我们研究的目的是确定先前治疗线的数量如何影响 ASCT 和 CAR-T 患者的结果。

对 2016 年 6 月至 2020 年 4 月在罗切斯特梅奥诊所接受 ASCT 和/或 axicabtagene ciloleucel (axi-cel) 治疗 R/R 侵袭性 LBCL 的患者进行了回顾性研究。该研究是根据赫尔辛基宣言进行的，并获得了梅奥诊所 IRB 的批准。对患者进行管理，包括根据机构指南和治疗医师的判断决定桥接治疗以及 ASCT 前和 ASCT 后或 CAR-T 成像的时间。关于基线特征评估、治疗线定义和反应评估的方法的其他详细信息在补充中提供。组间的不同比较基于在现实世界中遇到的 ASCT 或 CAR-T 之前的治疗线（图 S1，联合图）。患者在两线（ASCT [2]）或三线（ASCT [3]）化疗后接受了 ASCT。对于 CAR-T 组，最早可能的适应症患者接受 CAR-T（CAR-T[early]）是在两线化疗后（CAR-T [2]）和/或在 ASCT [2] 之后。那些不符合 CAR-T[早期] 标准的被归类为 CAR-T[晚期]。此外，患者可以在 ASCT 复发（CAR-T[ASCT]）后立即接受 CAR-T，无需任何间隔治疗。该组包括来自 ASCT [2] 和 ASCT [3] 组的患者。那些不符合 CAR-T[ASCT] 标准的被归类为 CAR-T[other]，因为这些需要在 ASCT 和 CAR-T 之间进行额外的治疗。所有分析均在 R3.6.3 中进行，患者接受 CAR-T（CAR-T[早期]）的最早可能适应症是在两线化疗后（CAR-T [2]）和/或 ASCT [2] 后。那些不符合 CAR-T[早期] 标准的被归类为 CAR-T[晚期]。此外，患者可以在 ASCT 复发（CAR-T[ASCT]）后立即接受 CAR-T，无需任何间隔治疗。该组包括来自 ASCT [2] 和 ASCT [3] 组的患者。那些不符合 CAR-T[ASCT] 标准的被归类为 CAR-T[other]，因为这些需要在 ASCT 和 CAR-T 之间进行额外的治疗。所有分析均在 R3.6.3 中进行，患者接受 CAR-T（CAR-T[早期]）的最早可能适应症是在两线化疗后（CAR-T [2]）和/或 ASCT [2] 后。那些不符合 CAR-T[早期] 标准的被归类为 CAR-T[晚期]。此外，患者可以在 ASCT 复发（CAR-T[ASCT]）后立即接受 CAR-T，无需任何间隔治疗。该组包括来自 ASCT [2] 和 ASCT [3] 组的患者。那些不符合 CAR-T[ASCT] 标准的被归类为 CAR-T[other]，因为这些需要在 ASCT 和 CAR-T 之间进行额外的治疗。所有分析均在 R3.6.3 中进行，患者可以在 ASCT 复发后立即接受 CAR-T（CAR-T[ASCT]），无需任何间隔治疗。该组包括来自 ASCT [2] 和 ASCT [3] 组的患者。那些不符合 CAR-T[ASCT] 标准的被归类为 CAR-T[other]，因为这些需要在 ASCT 和 CAR-T 之间进行额外的治疗。所有分析均在 R3.6.3 中进行，患者可以在 ASCT 复发后立即接受 CAR-T（CAR-T[ASCT]），无需任何间隔治疗。该组包括来自 ASCT [2] 和 ASCT [3] 组的患者。那些不符合 CAR-T[ASCT] 标准的被归类为 CAR-T[other]，因为这些需要在 ASCT 和 CAR-T 之间进行额外的治疗。所有分析均在 R3.6.3 中进行，p值是两侧的，显着性水平设置为 <0.05。

研究期间共有 105 名患者接受了 ASCT，62 名患者接受了 CAR-T 治疗。在 ASCT 患者中，87 人在两线化疗后接受了 ASCT（ASCT [2]），18 人在三线化疗后接受了 ASCT（ASCT [3]）。整个 ASCT 和 CAR-T 队列的基线特征如表 S1 所示。ASCT[2]和ASCT[3]组的比较见表S2。ASCT 组的 26 名患者（25%）（n = 19，ASCT [2] 和 n = 7，ASCT [3]）接受了 CAR-T 作为后续治疗。除了 ASCT [3] 组在一线治疗后疾病进展的患者百分比较高（44% vs 21%，p < 0.001) 和二线治疗。中位随访时间为 19.7 个月，ASCT [2] 组的 1 年无事件生存 (EFS) 率显着高于 ASCT [2] 组 67%，(95%CI 58%–78%) 与 ASCT [3] 44 %，（95% CI 27%–75%；p  = 0.015），图 1（A）。然而，1 年总生存 (OS) 率没有显着差异（图 1（B））。

在 CAR-T 接受者（n = 62）中，28 名（45%）患者曾接受过 ASCT。只有 25/62 (40%) 患者在最早可能的适应症（CAR-T[early]）接受了 CAR-T，其中 15 名患者属于 CAR-T [2] 组，10 名患者属于 ASCT [2]组（表 1）。CAR-T[早期] 与 CAR-T[晚期] 组的基线特征具有可比性。在 CAR-T 队列中位随访 12.1 个月时，1 年 EFS 有改善的趋势（48%，（95%CI 32–72） vs CAR-T [晚期] 30%，（95 %CI 18%–49%) p  = 0.055) 和 OS (75%, (95%CI 60–95) vs CAR-T [晚期] 56%, (95%CI 42%–76%) p = 0.053) 在 CAR-T[早期] 组中（图 1（C）和（D））。接下来，我们比较了 CAR-T[ASCT]（n = 14、10 - ASCT [2] 和四个 - ASCT [3]）和 CAR-T[其他]（n = 48）组。CAR-T[other] 组中显着更高的患者比例（75% vs 43%，p  = 0.03）需要在 CAR-T 输注前进行桥接治疗（表 1）。CAR-T[ASCT] 组的 1 年 OS 率为 77%（95%CI 60-100），而 CAR-T[other] 组为 60%（95%CI 47-76）。p  = 0.039)，图 1(E) 和 (F)。最后，对ASCT[3]组（n=18）和CAR-T[2]组（n=15）进行了分析，两组具有可比性（表S3），除了CAR-T[2]组的患者明显更多疾病进展至二线化疗（87% vs 44%，p = 0.01)，并且从诊断到输注的中位时间更短。在中位随访 9.9 个月时，两组之间的 1 年 EFS 或 OS 率没有显着差异（图 1（G）和（H））。

本报告根据现实世界中临床上遇到的先前疗法，在一个大型学术中心提供了 ASCT 和 CAR-T 的临床结果。多项研究试图确定可以预测 CAR-T 持久反应的患者、疾病和产品相关因素，以便更好地选择患者。我们试图根据临床环境中遇到的先前治疗线来确定何时最佳地纳入 CAR-T，而不是根据 FDA 标签的指示。我们发现 ASCT [2] 与 ASCT [3] 患者的 1 年 EFS 率更长（67% 对 44%）。CORAL 试验还证明了后续治疗线的 EFS 和 OS 逐渐缩短。然而，我们没有发现 ASCT [2] 和 ASCT [3] 组之间 1 年 OS 率的差异，可能是由于 ASCT [3] 组中的人数少和后续治疗的使用 (7/18)。此外，ASCT [3] 组还包括患有更原发性难治性疾病的患者，该组已知预后较差。CAR-T（早期）组显示出更好的 1 年 EFS 和 OS 的趋势，CAR-T[ASCT] 组的 OS 显着改善，这可能表明更好的疾病生物学，因为进行 ASCT 需要化学敏感性。这也很明显，因为 CAR-T（其他）组对桥接治疗的需求增加。塞尔默等人。之前已经将 CAR-T 的结果与两个先前治疗线以外的替代疗法进行了比较，并显示连续治疗线的 CAR-T 结果恶化。ASCT [3] 组还包括患有更原发性难治性疾病的患者，该组已知预后较差。CAR-T（早期）组显示出更好的 1 年 EFS 和 OS 的趋势，CAR-T[ASCT] 组的 OS 显着改善，这可能表明更好的疾病生物学，因为进行 ASCT 需要化学敏感性。这也很明显，因为 CAR-T（其他）组对桥接治疗的需求增加。塞尔默等人。之前已经将 CAR-T 的结果与两个先前治疗线以外的替代疗法进行了比较，并显示连续治疗线的 CAR-T 结果恶化。ASCT [3] 组还包括患有更原发性难治性疾病的患者，该组已知预后较差。CAR-T（早期）组显示出更好的 1 年 EFS 和 OS 的趋势，CAR-T[ASCT] 组的 OS 显着改善，这可能表明更好的疾病生物学，因为进行 ASCT 需要化学敏感性。这也很明显，因为 CAR-T（其他）组对桥接治疗的需求增加。塞尔默等人。之前已经将 CAR-T 的结果与两个先前治疗线以外的替代疗法进行了比较，并显示连续治疗线的 CAR-T 结果恶化。CAR-T（早期）组显示出更好的 1 年 EFS 和 OS 的趋势，CAR-T[ASCT] 组的 OS 显着改善，这可能表明更好的疾病生物学，因为进行 ASCT 需要化学敏感性。这也很明显，因为 CAR-T（其他）组对桥接治疗的需求增加。塞尔默等人。之前已经将 CAR-T 的结果与两个先前治疗线以外的替代疗法进行了比较，并显示连续治疗线的 CAR-T 结果恶化。CAR-T（早期）组显示出更好的 1 年 EFS 和 OS 的趋势，CAR-T[ASCT] 组的 OS 显着改善，这可能表明更好的疾病生物学，因为进行 ASCT 需要化学敏感性。这也很明显，因为 CAR-T（其他）组对桥接治疗的需求增加。塞尔默等人。之前已经将 CAR-T 的结果与两个先前治疗线以外的替代疗法进行了比较，并显示连续治疗线的 CAR-T 结果恶化。⁷然而，研究中的替代疗法组由长期（2001-2017 年）接受治疗的患者组成，包括除 ASCT 以外的治疗。Locke 等人最近的一项分析。确定较高的肿瘤负荷、较低的 CAR-T 细胞扩增、促炎性肿瘤微环境和增加的 CAR-T 细胞倍增时间是与缺乏持久反应相关的因素。这些因素随着治疗线数量的增加而显着显着，并支持了 CAR-T 疗效随着治疗线增加而降低的观察结果。⁸连续治疗线的更糟糕结果可能是由于疾病难治性增加、高风险疾病特征和更差的 CAR-T 细胞产品适应性的组合。这些和其他因素中的每一个导致恶化结果的程度还有待确定。根据 FDA 的标签，我们的队列中只有 40% 的患者（25/62）在最早可能的适应症下转诊到 CAR-T。因此，早期转诊是改善 CAR-T 结果以避免后勤延误和桥接治疗需求的关键。对于首次抢救后未达到 CR 的患者，是尝试再次抢救还是继续使用 CAR-T 仍然没有答案。目前正在朝这个方向进行大型注册研究。我们将 ASCT(3) 与 CAR-T(2) 进行比较的分析未发现 1 年 EFS 或 OS 率存在差异。这可能是由于 CAR-T(2) 组中化疗难治性疾病占二线治疗的比例较高，从诊断到输注的时间较短，以及队列中的总体患者人数较少。该研究受限于其回顾性、患者人数少和中位随访时间短，这可能与所进行的比较中缺乏生存差异有关。然而，在同一时期比较 ASCT 和 CAR-T 队列避免了支持性护理差异的影响，单中心分析提供了一个在实践模式方面更同质的队列。需要具有更长随访时间的更大队列来了解这些治疗方式排序的含义。