CAR T-cell therapy for CNS R/R ALL: A post hoc analysis of pooled data from five clinical trials

Chimeric antigen receptor (CAR) T-cell therapy has improved treatment for young patients with relapsed or refractory (R/R) B-cell acute lymphoblastic leukemia (ALL), with 1-year relapse-free survival rates reaching 60%. However, central nervous system (CNS) involvement is seen in 20% of all relapses; CNS disease is associated with a reduced overall survival (OS) after relapse and is difficult to treat. Safety and efficacy data of CD19-directed CAR T-cell therapy for CNS relapse remains scarce, as most trials have excluded patients with active CNS disease. Therefore, Leahy et al. performed a post hoc analysis that evaluated the safety and activity of CAR T-cell therapy in patients with a history of CNS R/R B-cell ALL.¹

Methods

This study included a total of 195 patients with R/R CD19-positive ALL or lymphocytic lymphoma from five clinical trials (Pedi CART19 [NCT01626495], 13BT022 [NCT02374333], 16CT022 [NCT02906371], ENSIGN [NCT02228096], and ELIANA [NCT02435849]) from the Children’s Hospital of Philadelphia, Philadelphia, US. The participants of these studies received CD19-directed CAR T-cell therapy between April 17, 2012, and April 16, 2019.

A total of 154 patients from the Pedi CART19, ELIANA, ENSIGN, and 16CT022 trials received tisagenlecleucel, and 41 patients from the 13BT022 trial received huCART19 (humanized CD19-directed CAR-T cells).

Patients were included if they had a positive control of CNS disease at enrolment and infusion; and patients were excluded if they required treatment for acute neurological toxic effects (greater than Grade I in severity) or had parenchymal lesions that were likely to increase the risk of neurotoxicity.

Patients were categorized according to CNS status at relapse or within the 12 months preceding CAR T-cell infusion—either CNS-positive or CNS-negative disease (Table 1).

Patients with CNS-positive disease were further subcategorized based on the morphological bone marrow involvement—defined as either combined bone marrow and CNS involvement, or isolated CNS involvement.

Endpoints were the number of patients with complete response 28 days following infusion, incidence of cytokine release syndrome and neurotoxicity in the first 8 weeks after infusion, and Kaplan-Meier analysis of relapse-free survival (RFS) and OS.

Table 1. Baseline characteristics by subgroup*

BM, bone marrow; CAR, chimeric antigen receptor; CNS, central nervous system; CSF, cerebrospinal fluid; HSCT, hematopoietic stem cell transplant; IQR, interquartile range; neg, negative; pos, positive; WBC, white blood cells. *Adapted from Leahy et al.1 †Defined at relapse or refractory evaluation. ‡Includes patients in whom pre-infusion bone marrow evaluation and lumbar puncture were done at enrolment (3–6 weeks before CAR T-cell infusion) and who received bridging chemotherapy in the interim period. §Two patients in the isolated CNS stratum developed marrow disease during the bridging period. ¶CNS1 = CSF with <5 WBC/μL and no blasts; CNS2 = CSF with <5 WBC/μL and positive for leukemic blasts; and CNS3 = CSF with ≥5 WBC/μL and positive for blasts or parenchymal or cranial nerve involvement. #These four patients with CNS3 pre-infusion showed residual enhancement by magnetic resonance imaging (optic nerve [n=1]; cranial nerve V, cranial nerve VII, and cerebellar lesion [n=1]; or cauda equina nerve roots [n=1]) or leukemic blasts in the CSF (six WBC/µL [n=1]).
Characteristic, n (%) (unless otherwise specified)	All patients (N = 195)	CNS stratification			CNS categorization by BM involvement
		CNS-neg (n = 129)	CNS-pos (n = 66)	p value	BM or combined BM and CNS involvement (n = 152)	Isolated CNS involvement† (n = 43)	p value
Median age at infusion, years (IQR)	11.3 (7.7–16.5)	12.4 (7.6–17.0)	10.2 (7.9–15.0)	0.63	12.3 (7.8–17.0)	10.2 (8.0–13.7)	0.35
Age group, years
<3	8 (4)	4 (3)	4 (6)	0.76	4 (3)	4 (9)	0.29
3 to <10	72 (37)	47 (36)	25 (38)	—	57 (38)	15 (35)	—
10 to <18	82 (42)	56 (43)	26 (39)	—	64 (42)	18 (42)	—
≥18	33 (17)	22 (17)	11 (17)	—	27 (18)	6 (14)	—
Sex				0.94			0.80
Female	85 (44)	56 (43)	29 (44)	—	67 (44)	18 (42)	—
Previous HSCT	94 (48)	59 (46)	35 (53)	0.33	73 (48)	21 (49)	0.93
Disease status at referral				<0.0001			0.0002
Primary refractory	27 (14)	26 (20)	1 (2)	—	27 (18)	0 (0)	—
First relapse	59 (30)	44 (34)	15 (23)	—	50 (33)	9 (21)	—
Second or greater relapse	109 (56)	59 (46)	50 (76)	—	75 (49)	34 (79)	—
Neurological comorbidity
Stroke	4 (2)	3 (2)	1 (2)	>0.99	3 (2)	1 (2)	>0.99
Seizure	26 (13)	14 (11)	12 (18)	0.15	18 (12)	8 (19)	0.25
Methotrexate toxicity	15 (8)	11 (9)	4 (6)	0.54	12 (8)	3 (7)	>0.99
Neurological deficit	13 (7)	7 (5)	6 (9)	0.37	8 (5)	5 (12)	0.17
BM disease burden before infusion				<0.0001			<0.0001
<0.01%	74 (38)	30 (23)	44 (67)	—	39 (26)	35 (81)	—
0.01–4.99% (M1)	30 (15)	22 (17)	8 (12)	—	24 (16)	6 (14)	—
5–25% (M2)	19 (10)	17 (13)‡	2 (3)	—	19 (13)	0 (0)	—
>25% (M3)	72 (37)	60 (47)‡	12 (18)‡	—	70 (46)	2 (5)§	—
CNS status before infusion¶				0.021			0.054
CNS1	177 (91)	120 (93)‡	57 (86)‡	—	139 (91)	38 (88)	—
CNS2	14 (7)	9 (7)‡	5 (8)	—	12 (8)	2 (5)	—
CNS3#	4 (2)	0 (0)	4 (6)	—	1 (1)	3 (7)	—
Seizure prophylaxis	64 (33)	22 (17)	42 (64)	<0.0001	33 (22)	31 (72)	<0.0001

Results

• The median length of follow-up was 39 months (interquartile range, 25–49) in the CNS-positive subgroup and 36 months (interquartile range, 18–49) in the CNS-negative subgroup.
• Of the 195 patients, 34% were categorized as CNS-positive, 66% as CNS-negative, and 22% as isolated CNS involvement.
• The number of patients with a complete response at 28 days following infusion was similar between the CNS-positive and CNS-negative subgroups (97% vs 94%, respectively; p = 0.74; Table 2).
• There was no statistically significant difference in RFS or OS between the CNS-positive and CNS-negative subgroups, respectively.
  • RFS at 2 years: 60% (95% confidence interval [CI], 49–74) vs 60% (95% CI, 51–71); p = 0.50
  • OS at 2 years: 83% (95% CI, 75–93) vs 71% (95% CI, 64–79); p = 0.39
• OS at 2 years was significantly greater for patients with isolated CNS involvement compared with those with bone marrow involvement (91% [95% CI, 82–100] vs 71% [95% CI, 64–78], respectively; p = 0.046).
• The only factor independently associated with higher risk of relapse after CAR T-cell therapy was a high disease burden (M3) in the bone marrow pre-infusion (multivariate analysis hazard ratio, 5.354 [95% CI, 2.778–10.321]; p < 0.0001).

Table 2. Disease outcomes following CAR T-cell therapy*

BM, bone marrow; CAR, chimeric antigen receptor; CNS, central nervous system; CSF, cerebrospinal fluid; IQR, interquartile range; neg, negative; pos, positive; WBC, white blood cells. *Adapted from Leahy et al.1 †Three patients died before evaluation at Day 28 from sequelae of cytokine release syndrome, coagulopathy, and infection without clear evidence of progressive disease. ‡CNS1 = CSF with <5 WBC/μL and no blasts; CNS2 = CSF with <5 WBC/μL and positive for leukemic blasts; and CNS3 = CSF with ≥5 WBC/μL and positive for blasts or parenchymal or cranial nerve involvement.
Outcome	All patients (N = 195)	CNS stratification			CNS categorization by BM involvement
		CNS-neg (n = 129)	CNS-pos (n = 66)	p value	BM or combined BM and CNS involvement (n = 152)	Patients with isolated CNS involvement (n = 43)	p value
Disease response at Day 28, n (%)
Complete response	185 (95)	121 (94)	64 (97)	0.74	143 (98)	42 (98)	0.35
No response	7 (4)	6 (5)	1 (2)		7 (5)	0 (0)
Not evaluable†	3 (2)	2 (2)	1 (2)		2 (1)	1 (2)
Patients with relapse, n/N (%)	72/185 (39)	45/121 (37)	27/64 (42)	0.51	56/143 (39)	16/42 (38)	0.90
CNS status at relapse, n/N (%)‡				0.0066			0.0062
CNS1	45/72 (63)	33/45 (73)	12/27 (44)		38/56 (68)	7/16 (44)
CNS2	4/72 (6)	1/45 (2)	3/27 (12)		1/56 (2)	3/16 (19)
CNS3	7/72 (10)	1/45 (2)	6/27 (22)		3/56 (5)	4/16 (25)
Unknown	16/72 (22)	10/45 (22)	6/27 (22)		14/56 (25)	2/16 (12)
Median follow-up (IQR), months	37 (21–49)	36 (18–49)	39 (25–49)	0.73	36 (18–49)	40 (29–52)	0.38

Safety

There was no difference in the incidence or severity of cytokine release syndrome or neurotoxicity between the CNS-negative and the CNS-positive disease subgroups (Table 3).

Table 3. Cytokine release syndrome and neurotoxic adverse events by CNS status*^†

CNS, central nervous system. *Adapted from Leahy et al.1 †No Grade 5 adverse events were reported.
Adverse event	CNS-negative disease (n = 129)	CNS-positive disease (n = 66)	p value
Cytokine release syndrome, %			0.26
Grade 0	15	20
Grade 1	9	3
Grade 2	47	58
Grade 3	14	11
Grade 4	15	9
Any neurotoxicity, %			0.20
Grade 0	59	41
Grade 1	19	30
Grade 2	11	15
Grade 3	9	9
Grade 4	2	3

Conclusion

In summary, the two CAR-T cell therapies, tisagenlecleucel and huCART19, show promising efficacy in reducing CNS disease and maintaining durable remissions in young patients with CNS R/R B-cell ALL or lymphocytic lymphoma. As CNS involvement in the multiple-relapsed setting, particularly following radiation and hematopoietic stem cell transplantation, is difficult to treat and has a poor outcome, the durable remissions observed in this patient population represent a major advance in the treatment of patients with CNS relapse ALL. The overall incidence of CNS relapse was low, and the results indicate that patients with CNS disease that is adequately controlled before infusion can benefit from CD19 CAR T-cells, without increasing the risk of severe neurotoxicity.