Cellular immunotherapy for refractory DLBCL in the CART era: still a role for allogeneic transplantation?

Abstract

Chimeric antigen receptor-engineered T (CART) cells are a promising new treatment option for patients with multiply relapsed and refractory (R/R) diffuse large B-cell lymphoma (DLBCL). Because of the favourable outcome data reported for CART cells, uncertainty is emerging if there is still a role for allogeneic hematopoietic cell transplantation (alloHCT) in the treatment of R/R DLBCL. This paper provides an overview of available evidence and theoretical considerations to put these two types of cellular immunotherapy (CI) into perspective. Altogether, current data suggest that CART cells are preferred now over transplantation as first choice CI in many clinical situations. However, the majority of patients will fail CART therapy, resulting in an unmet medical need where alloHCT could be beneficial. In contrast, employing alloHCT instead of CART cells as 1st CI should be presently restricted to situations where CART cell therapy deems not feasible or useful, such as patients with refractory cytopenia or incipient MDS. However, alloHCT remains a standard treatment option as 1^st CI for patients with in chemosensitive R/R DLBCL when CARTs are not available, or transplantation is preferred by the patient. Continuous collection and analysis of CI outcome data by professional registries appear to be of key importance for developing rational strategies of CI allocation and sequencing.

Allogeneic hematopoietic cell transplantation (alloHCT) is a potentially curative cellular immunotherapy (CI) for relapsed or refractory (R/R) diffuse large B-cell lymphoma (DLBCL), and used to be considered as a standard therapeutic option in this setting for eligible patients (1–4). However, the recent approval of CD19-directed chimeric antigen receptor-engineered T cells (CART cells) for treatment of patients with DLBCL having failed two lines of therapy has added a promising new immunotherapeutic tool to the treatment armory of R/R DLBCL. Because of the favourable outcome data reported for CART cells, uncertainty is emerging if there is still a role for alloHCT in the management of multiply R/R DLBCL. As a result, in some recently proposed treatment algorithms, alloHCT has completely disappeared (5).

While CART cell treatment is increasingly used as first CI in patients with multiply R/R DLBCL, the aim of this effort was to identify settings where alloHCT promises to be beneficial in this new treatment landscape. Based on a comprehensive summary of mode of action, efficacy, toxicity, outcome predictors, rescue options, and strengths and limitations of both modalities, a Task Force appointed by the EBMT Lymphoma Working Party and the CIBMTR Lymphoma Working Group (see Supplements for details) developed this proposal for a rational approach to alloHCT at the dawn of the CART era when valid evidence is still sparse.

Scientific basis / current evidence

Mode of action

CART cells:

In contrast to alloHCT, where the immunotherapeutic effect relies on a polyclonal immune reaction against multiple undefined target antigens (graft-versus-lymphoma activity; GVL), CART cells exert monoclonal immune activity against defined antigens, such as CD19, thereby avoiding the unspecific graft-versus-host reactions linked to alloHCT efficacy. This immune effect is mediated by the chimeric antigen receptor (CAR) unique to CART cells, thereby combining the HLA-independent antigen-binding properties of the B cell receptor with the potent activating and effector functions inherent to T cells (6, 7).

An essential part of CART therapies is lymphodepleting conditioning which is administered to the patient prior to infusion of the CART product (8–10). Although not fully understood, this lymphodepletion enhances CART cell expansion and has been associated with efficacy, possibly by promoting a pro-inflammatory cytokine milieu, eradicating regulatory T cells, abrogating endogenous T cell responses against the transgene product, and/or simply creating space for the re-infused cells to expand (6, 7, 11, 12).

AlloHCT:

Similar to other lymphoid malignancies, the basis of alloHCT in DLBCL is graft-versus-lymphoma activity (GVL). Evidence for GVL efficacy in DLBCL derives from observations showing that (1) alloHCT but not autoHCT can overcome pretransplant metabolic chemoresistance (13–15), (2) reduced-intensity conditioning (RIC) alloHCT can provide long-term disease control in patients who have failed myeloablative treatment with autoHCT(16–21), (3) T-cell depletion of the graft has detrimental effects on the relapse risk (22) and (4) withdrawal of immune suppression or donor lymphocyte infusion can induce remissions in DLBCL patients progressing after alloHCT (23). Similar to CART cells, successful engraftment of allogeneic hematopoietic cells requires immunosuppressive conditioning (24, 25). Another aim of conditioning in rapidly proliferating neoplasms is to control the disease until the immunotherapeutic potential of the donor immune system can become effective after immunosuppression tapering, i.e. bridging to GVL. This is the rationale for using intensive conditioning in DLBCL (26). However, the superiority of myeloablative conditioning over reduced-intensity conditioning (RIC) in DLBCL is not supported by retrospective studies (3, 16–18, 25, 27), again arguing that the GVL effects are key to disease eradication in DLBCL.

Efficacy

CART cells:

The approval trials for both labeled constructs (Axicabtagene ciloleucel, Axicel; Tisagenlecleucel, Tisacel) were largely restricted to patients with DLBCL who had failed >=2 lines of chemoimmunotherapy and/or autoHCT and had active disease at the time of enrolment. Overall response rates (ORR) for those patients who were actually infused with the CART product were consistently high with 52%−83% including 40%−55% complete responses (CR). Whereas partial responses were largely transient, the majority of CRs were sustained, resulting in 12-month progression-free survival (PFS) and overall survival (OS) rates of 34%−44% and 49%−59%, respectively(8, 28, 29) (Table 1a). Considerable differences regarding the turnaround time between apheresis and reinfusion existed, partly explaining why outcomes differed between the trials when calculated by intent-to-treat (ITT)(ORR 75% and 30%, for Axicel and Tisacel, respectively)(30). Preliminary “real-world” experience with commercial CART therapy roughly confirms the response data from the approval trials (31–35)(Table 1b).

Table 1a.

Results of clinically approved or far developed CD19 CARTs in DLBCL: Prospective trials

	ZUMA-1 NCT02348216 (8,28)	JULIET NCT02445248 (29)	TRANSCEND NHL-001 NCT02631044 (83)
Study type	Phase 1/2 prospective	Phase 2 prospective	Phase 1 prospective
Indication	DLBCL or PMBCL or tFL; PD/SD and/or autoHCT failure <12mo	DLBCL; failed 2 lines and/or autoHCT	DLBCL, PMBCL, FL3B; failed 2 lines
N (enrolled/infused)	119 / 108	165 / 115	268/342
Data reported using intent-to- treat principle?	No	No	No
Age (years; median (range)	58 (23–76)	56 (22–76)	63 (18–86)
PS >1 (ECOG)	0%	0%.	n.a.
Prior autoHCT	21%	49%	34%*
Disease status refractory	79%	55%	67%
Construct (Costimulatory domain)	Axicabtagene ciloleucel (CD28)	Tisagenlecleucel (4–1BB)	Lisocabtagene maraleucel (4–1BB)
Bridging therapy permitted* (% patients administered)	No	Yes (90%)	Yes (56%)
Median turnaround time between leukapheresis and reinfusion (d)	17	54	24
Lymphodepleting regimen	FC	FC or bendamustine	FC
Response rate of infused patients (ORR/CR)	83% / 58%	52% / 38%	73%/53%
ORR by ITT	75%	36%	57%
Non-relapse mortality
	3%	0	1.5%
Progression-free survival (of infused patients)
1y	44%	35%	6.8 mo median
2y	38%
Overall survival (of infused patients)
1y	59%	49%	19.9 mo median
1.5y	52%	40%
2y	51%
Risk factors for response	Tumor load	Need for bridging	none
Grade ≥3 neutropenia (overall/> d +28)	80%/26%	33%/25%	60%/n.a.§
Grade ≥3 thrombopenia (overall/> d +28)	40%/24%	28%/39%	27%/n.a.
CRS (grade ≥3)	11%¶	22%‡	2%
Grade ≥3 neurotoxicity	32%	11%	10%
Follow-up (months)	27 (IQR 26–29)	14 (0–26)	11

^*salvage therapy before leukapheresis or between leukapheresis and infusion

^§37% had grade ≥3 cytopenia beyond day 28

^¶by Lee criteria (36)

^‡by Penn criteria (36)

CR, complete response; FC, fludarabine/cyclophosphamide; ITT, intent-to-treat; NA, not available; ORR, overall response rate.

Table 1b.

Results of clinically approved CD19 CARTs in DLBCL: Real-World data

	US Academic center ”Real World” (31)	Boston/Seattle ”Real World” (32)	CIBMTR Axi-cel (33)	US Academic centers (34)	US Academic centers (34)	CIBMTR Tisa-cel (35)
Study type	Retrospective	Retrospective	Retrospective	Retrospective	Retrospective	Retrospective
Indication	DLBCL,PMBCL,tFL	DLBCL,PMBCL,tFL, DLBCL transformed from non-FL	DLBCL or PMBCL or transforme d lymphoma	DLBCL, PMBCL, tFL, HGBL	DLBCL, tFL, HGBL	DLBCL or transformed lymphoma
N (enrolled/infus ed)	211 / 165	87 / 76	n.a./ 295	163/149	79/75	70
Data reported using intent-to-treat principle?	NA	NA	No	No	No	No
Age (years;median (range)	59 (21–82)	64	61 (19–81)	58 (18–85)	67 (36–88)	65 (19–89)
PS >1 (ECOG)	19%	n.a.	5%	14%.	6%	4%.
Prior HCT (auto/allo)	31%/NA	≈30%	34%/NA	29%/NA	23%/NA	23%/6%
Disease status refractory	NA	NA	66%	NA	NA	NA
Construct	Axicabtagene ciloleucel	Axicabtagene ciloleucel	Axicabtagene ciloleucel	Axicabtagene ciloleucel	Tisagenlecleucel	Tisagenlecleucel
Bridging therapy permitted* (%patients administered)	Yes	Yes (40%)	NA	Yes (61%)	Yes (72%)	Yes
Median turnaround time between leukapheresis and reinfusion (d)	27	n.a.	NA	28	44	NA
Lymphodepleting regimen	FC	FC	FC	FC	FC	FC
Response rate of infused patients (ORR/CR)	79% / 50%	64% / 41%	70% / 52%	72% / 43%$	59%/44%$	59% / 38%
ORR by ITT	62%	56%	NA	NA	NA	NA
Non-relapse mortality	2%	7%	NA	8%	6%	0
Progressionfree survival (of infused patients)	NA	NA	NA	NA	NA	NA
1y
2y
Overall survival				NA
(of infused patients)	n.a.	n.a.	NA		NA	NA
1y
1.5y
2y
Risk factors for response	ECOG >1; bulk; refractory disease; male sex	ECOG >1; bulk; IPI >2; CRP; prior ibrutinib	NA	NA	NA	NA
Grade ≥3 neutropenia	n.a.	n.a.	NA/ 7%§	NA	NA	NA
(overall/> d+28)
Grade ≥3 thrombopenia	n.a.	n.a.	NA/ 7%§	NA	NA	NA
(overall/> d+28)
CRS (grade ≥3)	7%	16%	10%	13%	1%	4%
Grade ≥3 neurotoxicity	33%	39%	NA	41%	3%	4%
Follow-up (months)	<3	4	6 (1–14)	NA	NA	6 (1–9)

^*salvage therapy before leukaoheresis or between leukapheresis and reinfusion

^§grade ≥3 cytopenia beyond day 28

^$d30 responses

CR, complete response; FC, fludarabine/cyclophosphamide; ITT, intent-to-treat; n.a., not available; ORR, overall response rate.

AlloHCT:

The only recent published prospective trial on alloHCT for R/R DLBCL is DSHNHL 2004-R3 (NCT00785330) for aggressive lymphoma, which enrolled 84 patients including 42 patients with DLBCL (26). Although the trial was formally following an ITT design, patients were eligible only after donor search was completed successfully, resulting in positive patient selection similar to the CART approval trials. With 55% of the patients being chemorefractory at alloHCT, PFS and OS were 45% and 52% at 1 year and 39% and 42% at 3 years post transplant. 1-year relapse/progression incidence was 29% with no relapse event occurring thereafter (26). These results are in keeping with registry studies, consistently showing 3-year PFS rates of 30%−40%(3, 16–18, 20) (Table 2). By design, these studies only included patients who actually received an alloHCT.

Table 2.

Studies on alloHCT for DLBCL

	EBMT 2011 (16)	GITMO 2012 (17)	CIBMTR 2013 (84) MAC/RIC	CIBMTR 2016 (3)	EBMT 2010-R-12 (18) MAC / RIC	CIBMTR/EBM T LY16-03 (20) MRD / MUD TCD+ / MUD TCD-/haplo	DSHNHL 2004-R3 NCT007853 30 (26)
Study type	Registry retrospective	Registry retrospective	Registry retrospective	Registry retrospective	Registry retrospective	Registry retrospective	Phase 2 Prospective
Indication	autoHCT failure	autoHCT failure	Refractory disease	autoHCT failure	DLBCL,alloHCT as 1st HCT (prior autoHCT excluded)	DLBCL, 1st alloHCT (prior autoHCT permitted)	PIF, early relapse, autoHCT failure
N	101	165	533 (incl.80 FL3)	503	132 / 98	1306 525 / 403 / 378/132	84*
Period	1997–2006	1995–2008	1998–2010	2000–2012	2002–2010	2008–2015	2004–2009
Age (years; median (range))	46 (19–66)	43 (16–65)	46 (19–66) / 53 (20-70)	52 (19–72)	42 (31–50) / 52 (43-57)	55 (19–75)	48 (38–57)
PS >1 (ECOG) or <90% (Karnofsk y)	2%	n.a.	n.a.	10% §	16% / 19%	38%/38%/43%/27%	8%§
Prior autoHCT	100%	100%	15% / 38%	100%	0%	55%/59%/61%/42%	46%
Disease status refractory	26%	33%	100%	21%	46% / 32%	21%/18%/17%/17%	55%
Unrelated donor	29%	35%	34% / 53%	50%	37% / 34%	0%/100%/100%/0%	57%
Conditioning regimen RIC or NMA	64%	70%	43%	100%	0% / 100%	100%	0%
Nonrelapse mortality	16%	n.a.	38% / 25%	12%	10% / 5%	n.a.	12%
100	25%			23%	20% / 16%	13%/21%/20%/16%	35%
d	28%		47% / 36%	30%
1y   3y			53% / 42%		23% / 17%	17%/26%/30%/22%
Relapse incidence			27% /		46% /	39%/33%/28%/3
1y	24%	n.a.	32%	33%	46%	4%	29%
3y	30%		28% / 35%	38%	47% / 51%	47%/38%/34%/4 1%	29%
Progressi on-free survival	52%	48%	26% / 32%	44%	34% / 38%	48%/46%/52%/50%	45%
1y	42%	34%		31%			39%
3y	23%	27%	19% / 23%	HR 2.04 (vs CR)	30% / 33%	37%/36%/37%/38%
3y			19% / 23%			n.a.
(if refractory at HCT)
Overall survival
1y	65%	55%	31% / 42%	54%	39% / 50%	65%/56%/63%/66%	52%
3y	52%	42%	19% / 28%	37%	32% / 40%	50%/43%/46%/46%	42%
Risk factors for PFS/OS	Time to rel after auto <12mo / =	Time to rel after auto <12mo / Refractory at HCT, MUD	DLBCL (vs FL3)	Refractory at HCT, poor PS, MAC / Refractory at HCT, poor PS, MAC, Time auto_allo <12mo	n.a.	Refractory at HCT, poor PS / Refractory at HCT, poor PS, age, HCT-CI	n.a. / refractory at HCT, no ATG, >4 lines of pretreatme nt, mismatche d donor
Grade 2-4 (3-4) acute GVHD	33% (n.a.)	29% (10%)	29% / 31%	36% (15%)	n.a.	32%/32%/42%/34% (11%/13%/19%/7%)	30% (n.a.)
Extensive chronic GVHD	17%	n.a.	n.a.	33%	n.a.	n.a.	33%-41% (3y)
1y
Follow-up (months)	36 (3–112)	39 (1–144)	35 / 30	55 (11–49)	36 (9–68)	49 (12–73)	48 (29–71)

^*includes 27% patients with T cell lymphoma

^§Karnofky <80%

ATG, anti-thymocyte globuline; DLBCL, diffuse large B-cell lymphoma; GVHD, graft-versus-host disease; HCT-CI, hematopoietic cell transplantation comorbidity index; MAC, myeloablative conditioning; MUD, matched unrelated donor, n.a., not available; PS, performance status; RIC, reduced intensity conditioning; TCD, T-cell depletion

Safety

CART cells:

The major toxicities of the CD19 constructs discussed here consist of cytokine release syndrome (CRS), immune effector cell-associated neurotoxicity syndrome (ICANS)(36), and cytopenias. CRS is associated with mild to severe constitutional symptoms and usually develops within the first week after infusion (37). In the pivotal trials, grade ≥3 CRS has been observed in 11% and 22% of patients receiving Axicel or Tisacel, respectively, but the different grading systems used have to be taken into account (Table 1a).

ICANS is characterized by impaired vigilance, apraxia, confusion and aphasia, which can evolve to severe central symptoms such as seizures and coma (38). Higher grade ICANS is observed in up to 40% of Axicel-treated patients (less after Tisacel), usually within the first 2 weeks after CART cell infusion (8, 9, 29, 31, 32) (Table 1a+b). Upon standard treatment with high-dose steroids ICANS is mostly transient with a median duration of 5–11 days but can follow a prolonged course in individual patients (39, 40), and can contribute to NRM after CART which has ranged thus far from 0 to 9% (41). More than 20% of CD19 CART-treated patients suffer from protracted or recurrent neutropenia and/or thrombocytopenia beyond 4 weeks after reinfusion (28, 29, 42). Persistent neutropenia, along with B cell aplasia and hypogammaglobulinemia, contributes to the increased risk of infections observed in patients treated with CART cells (43).

AlloHCT:

Although early mortality has substantially decreased in recent years following the introduction of lower-intensity conditioning regimens, NRM still can be as high as 30% at two years (Table 2). This is largely due to complications of acute and chronic graft-versus-host disease (GVHD). In fact, grade 3–4 acute GVHD can occur in 10–15% of patients allografted for DLBCL, and up to one third of surviving patients at least transiently suffers from quality-of-life-impairing chronic GVHD (3, 17, 26). Other complications which can contribute to mortality and morbidity after alloHCT include neutropenic and opportunistic infections, transplant-related microangiopathy, and veno-occlusive disease (44, 45).

Eligibility and outcome predictors

CART cells:

Preliminary data suggest that poor performance status (PS), comorbidity, bulky disease, high tumor load, and high-risk IPI could be associated with inferior outcome of CD19-directed CART therapy (31, 32, 46–48). In contrast, older age (65–75) and chemorefractory disease were not associated with significant outcome disadvantages in the studies available to date (28, 29, 49, 50), although refractory patients tended to do worse in the approval trials. Of note, the presence of a rearrangement of myc along with bcl2 and/or bcl6 rearrangements (double/triple hit lymphomas; DHL), known to confer a poor prognosis, does not seem to affect CD19 CART therapy efficacy (29, 31, 32). More recently, “Real-World” data suggested that patients receiving Axicel as standard of care had a significantly poorer outcome on multivariate analysis when bridging treatment was needed (which had been an exclusion criterion for ZUMA-1)(51). Similarly, need for bridging was associated with inferior outcome in the JULIET trial (52). Finally, molecular indicators of pretreatment tumor activity and/or early response may emerge as sensitive outcome predictors (53).

Because of a potentially increased risk of neurotoxicity, patients with CNS involvement were excluded from ZUMA-1 and JULIET. Preliminary experience from commercial application of Axicel and Tisacel in patients with secondary CNS lymphoma, however, does not support this concern, while response rates appear to be comparable to patients without CNS involvement (54, 55). With the advent of other CD19-directed targeted therapies (56), CD19 CART therapy efficacy might be compromised by CD19-negative R/R DLBCL, even though an effect of CD19 expression on the response rate was not observed in the pivotal trials(8, 29).

Taken together, although the “usual suspects” PS and tumor aggressiveness seem to play a role, valid predictors of response and response duration after CART therapy have not been proven with the limited body of evidence available to date. However, patients with poor marrow function do not appear to be good candidates for successful CART therapy given that these were excluded from the pivotal trials but may be particularly susceptible to CART-associated cytopenias. This even more if impaired hematopoietic function is associated with morphological or clonal marrow abnormalities suspicious of impending or manifest myelodysplasia.

AlloHCT:

Patient-related factors associated with an increased risk of NRM after alloHCT for lymphoma include age 65 years or older, comorbidities, and resistant disease (57–60). The most consistent predictors of poor disease control after alloHCT for lymphoma appear to be refractory disease at transplant and short remission duration after a preceding autoHCT. Based on a study including 503 patients allografted for DLBCL recurrence after autoHCT, investigators from the CIBMTR proposed a score considering disease status at HCT, interval from autoHCT to alloHCT, and PS. Three-year PFS of patients with low risk (chemosensitive disease, good PS, and >12 months between auto- and alloHCT) was 40%, whereas it was <10% in patients meeting none of these criteria (3). In addition, a (mismatched) unrelated donor emerged as significant adverse predictor of PFS in larger cross-sectional studies of the main NHL subtypes (57, 59, 60). Only limited data exist on the efficacy of alloHCT in DHL (61).

Rescue options in case of failure

CART cells:

First real-world data obtained in patients with DLBCL progressing or relapsing after CART treatment indicate that the prognosis of these patients is poor, independently of the rescue strategy used (2^nd CART infusion, targeted therapy, chemotherapy), in particular if progression occurs within the first 3 months after dosing (62–64). Innovative strategies currently being explored include kinase inhibitors (65), polatuzumab (66), bispecific antibodies (67), lenalidomide (64), and, particularly promising, checkpoint inhibitors (63, 64, 68).

Since all of these agents will mostly provide only short-lived responses, investigating alloHCT consolidation for sensitive post-CART relapse in eligible patients appears to be worthwhile. However, it seems that in DLBCL this approach may often not be feasible if considered at all. In a real-world study, only 5 of 61 patients underwent alloHCT at any time post CART failure, with 2 of these patients being among 9 patients achieving sustained lymphoma control (62).

AlloHCT:

Although chemotherapy, immunomodulation, donor lymphocyte infusion, and second alloHCT could be considered for DLBCL recurrence post alloHCT, the efficacy of all these approaches appears to be limited (69–71), resulting in a generally poor prognosis of patients with DLBCL who progress after alloHCT (72). Preliminary data suggest that CART administration in this setting can be safe and effective, provided that there is no active GVHD and/or no need for immunosuppression (12, 73). At the University of Heidelberg, eight patients received CD19-directed CART therapy for lymphoma relapse after a preceding alloHCT. Safety and efficacy outcomes appeared to be comparable to that of similarly treated patients without prior alloHCT except for reactivation of chronic GVHD in 1 of 8 patients (ML Schubert, abstract submitted). Moreover, novel agents selectively targeting lymphoma cells while sparing GVL activity might improve the outlook of DLBCL relapse after alloHCT, alone or as a bridge to CART therapy (56, 66).

Strengths and limitations

CART cells:

The major strengths of CART (relative to alloHCT) are (a) an immediate anti-tumor effect upon infusion of the product, conferring immunotherapeutic activity even against actively proliferating DLBCL; (b) the avoidance of GVHD; and (c) a considerably lower NRM. The majority of CAR-T recipients relapse, but since CART treatment is only at the beginning of its evolution, it can be expected that further development of the methodology will likely improve treatment results of CART cells in DLBCL and other diseases (68, 74, 75).

Regarding limitations, it has to be taken into account that the population of patients included in published CART trials represent a favorable selection, e.g. by excluding patients with poor PS, cytopenias, CNS involvement, and those who deteriorate in the interval between apheresis and planned reinfusion (76). Due to the limited follow-up available to date, long-term safety and efficacy outcomes of CD19 CART therapies are still uncertain, and valid predictors of treatment success still need to be defined. Moreover, there is only very limited data on the efficacy of CD19 CARTs in the absence of active tumor, i.e. if used as consolidating treatment in patients with chemosensitive DLBCL (77). A major disadvantage of commercial CART therapies are their costs and the resulting limited availability in countries with less well equipped health systems.

AlloHCT:

Although mostly based on registry studies, available evidence regarding alloHCT in DLBCL appears to be robust in terms of case and study numbers, patient heterogeneity, risk factor characterization, and, in particular, long-term follow-up. Another strength of alloHCT is that it does not depend on a functioning patient T-cell system and can be readily performed in cytopenic patients.

On the other side, similar to CART trials, lymphoma alloHCT studies are positively biased because they are restricted to patients who were actually able to undergo transplantation, thereby disregarding transplant-ineligble patients and early progressors. A recent single-center analysis showed that only two thirds of all patients with DLBCL for whom a donor search was initiated finally were transplanted, resulting a long-term survival rate of 25% if calculated by intent-to-transplant (78). More importantly, despite significant improvement over time (79), alloHCT is still associated with substantial procedure-related mortality and morbidity. In addition, older age, comorbidities, and PS restrictions may affect alloHCT feasibility in patients with lymphoma. Furthermore, the limited efficacy of alloHCT in unresponsive DLBCL is another point to consider.

A rational approach to alloHCT in the CART era

The advent of CARTs has undoubtedly opened new perspectives for patients with R/R DLBCL. However, due to efficacy limitations and selection effects, the currently commercially available CARTs will address only a small portion of the unmet clinical need of patients with multiply relapsed DLBCL. AlloHCT thus should be actively explored as a rescue strategy for eligible patients failing CART cell therapy, both in clnical trials and by registry analyses. (Given the fact that in particular partial responses to CARTs are often not durable, a field of future research might be investigating alloHCT for consolidating incomplete DLBCL responses obtained with CART therapy, as currently being studied in ALL (80, 81). This approach could be refined by employing additional outcome predictors, such as circulating tumor DNA (53).) Hence, these two cellular immunotherapies could be complementary rather than mutually exclusive. Accordingly, pre-emptive donor search for all potentially transplant-eligible patients treated with CART cells appears worth considering. Bridging into alloHCT after CART failure remains a critical issue, but options for inducing remissions in chemorefactory DLBCL are steadily increasing (12, 64–67).

In patients with DLBCL who have failed two or more lines, both alloHCT and CARTs are available options. Although in the absence of comparative studies published survival rates do not clearly favor the currently approved CART cell therapies over alloHCT (Tables 1a & 2), treatment reality is that they are now the preferred option as first choice CI because of their safety and documented efficacy in refractory disease. Exceptions may consist in circumstances that limit the feasibility of CART therapy, such as refractory cytopenia with or without signs of therapy-related myelodysplasia, where considering alloHCT rather than CART therapy as first CI appears to be reasonable in eligible patients. Also autologous HCT (autoHCT) could still be an option beyond the 2^nd line (82). Since its successful application requires transplant eligibility, being autoHCT-naïve, and chemosensitivity, however, it will be a possibility only in selected cases.

In the absence of controlled studies, real-world outcome information addressing the effectiveness of CI approaches is crucial. The transplant field evolved utilizing real-world evidence through the use of outcomes databases for research. The contributions from the EBMT and CIBMTR registries have been invaluable to the field and helped to answer a multitude of questions to improve transplantation results. Similarly, EBMT and CIBMTR are now capturing outcomes of CART cell therapies allowing for quality control and research activities. Furthermore, collecting outcome data of transplants and CART therapies in the same registry appear to be of key importance for developing rational strategies of CI allocation and sequencing.

Conclusions

While CART cells are currently the preferred option as first CI for multiply R/R DLBCL, the natural role of alloHCT appears to be consolidation of sensitive relapses/ progressions after CART treatment in eligible patients who respond to (targeted) salvage therapy. Therefore using alloHCT as rescue strategy for CART failures should be actively explored in clinical trials and by registry analyses. Another potential field of research might be pre-emptive treatment of incomplete responses to CART cells with allotransplantation. In contrast, employing alloHCT instead of CART cells as 1st CI should be presently restricted to situations where CART cell therapy deems not feasible or useful, such as patients with refractory cytopenia or incipient MDS. However, alloHCT remains a standard treatment option as 1^st CI for patients with in chemosensitive R/R DLBCL when CARTs are not available, or transplantation is preferred by the patient.

Highlights.

• This paper addresses the possible role of alloHCT for R/R DLBCL in the CART era

• AlloHCT should be explored for patients failing CART therapy

• AlloHCT should be considered when CART therapy is not feasible or available

• AlloHCT and CART outcome data should be collected and analysed in registries