A revised definition for cure of childhood acute lymphoblastic leukemia.

With improved contemporary therapy, we reassess long-term outcome in patients completing treatment for childhood acute lymphoblastic leukemia (ALL) to determine when cure can be declared with a high degree of confidence. In six successive clinical trials between 1984 and 2007, 1291 (84.5%) patients completed all therapies in continuous complete remission. The post-therapy cumulative risk of relapse or development of a second neoplasm and the event-free survival rate and overall survival were analyzed according to the presenting features and the three treatment periods defined by relative outcome. Over the three treatment periods, there has been progressive increase in the rate of event-free survival (65.2% vs 74.8% vs 85.1% (P<0.001)) and overall survival (76.5% vs 81.1% vs 91.7% (P<0.001)) at 10 years. The most important predictor of outcome after completion of therapy was the type of treatment. In the most recent treatment period, which omitted the use of prophylactic cranial irradiation, the post-treatment cumulative risk of relapse was 6.4%, death in remission 1.5% and development of a second neoplasm 2.3% at 10 years, with all relapses except one occurring within 4 years of therapy. None of the 106 patients with the t(9;22)/BCR-ABL1, t(1;19)/TCF3-PBX1 or t(4;11)/MLL-AFF1 had relapsed after 2 years from completion of therapy. These findings demonstrate that with contemporary effective therapy that excludes cranial irradiation, approximately 6% of children with ALL may relapse after completion of treatment, and those who remain in remission at 4 years post treatment may be considered cured (that is, less than 1% chance of relapse).

INTRODUCTION

An objective definition of cure for children and adolescents with acute lymphoblastic leukemia (ALL) has been elusive. If eradication of the leukemic clone is the sole criterion, patients who maintain complete remission for a prolonged time after receiving all scheduled therapy may be considered “cured” of their disease. However, this definition does not take into account the possibility of a second cancer or other treatment-related complication that may jeopardize both the quality and length of the individual’s post-treatment survival.

In a previous study of 856 childhood ALL patients treated between 1962 and 1992 who were event-free survivors for 10 years after the induction of remission, we found that the cumulative risk of relapse at 30 years from the date of complete remission was extremely low (0.6%).[1] This finding was similar to that in another study of 1134 survivors who had been event-free for 10 years post-treatment at 20 years of follow-up.[2] However, we noted that among long-term survivors who had received radiation therapy, the development of second neoplasms continued well beyond 20 years of follow-up, without reaching a plateau,[1] a finding that has since been confirmed by other groups.[3-5] As a result, the irradiated group had a significantly lower survival rate than that of the non-irradiated 10-year event-free survivors, whose probability of survival was identical to that of the general population.[1,4] Based on these observations, we suggested a working definition of cure – 10 years or more of event-free survival – emphasizing that patients who meet this standard and did not receive radiation therapy have a normal survival expectancy.[1]

Improvement in the efficacy of treatment, together with the decreasing use of carcinogenic therapy, especially prophylactic cranial irradiation, has increased 5-year event-free survival rates to 80% to 87%.[6-10] This progress is likely to have an impact on long-term survival expectancy and may change the timing at which a child with ALL can be considered cured. We therefore re-examined the results of stopping therapy for patients enrolled in six consecutive clinical trials between the mid-1980s and the mid-2000s, a period in which the 5-year event-free survival rates increased from 68% to 87%.

MATERIALS AND METHODS

Study Population and Treatment Protocols

From February 1984 to October 2007, 1527 consecutive patients, 18 years or younger, with newly diagnosed ALL were enrolled in six successive clinical trials (Total Therapy studies 11, 12, 13A, 13B, 14 and 15) at St. Jude Children’s Research Hospital.[7,11-15] The clinical and biologic features of these patients are shown in the Supplementary Table 1, while the general characteristics of the different treatment protocols are summarized in the Supplementary Table 2. Briefly, prophylactic cranial irradiation was given to a decreasing proportion of patients in each successive trial because of improving systemic and intrathecal therapy: 64% in study 11, 37% in study 12, 22% in study 13A, 12% in study 13B and none in studies 14 and 15. In study 15, epipodophyllotoxins were omitted in all but 6% of the patients, who required intensification of therapy prior to hematopoietic stem cell transplantation. The treatment protocols were approved by the institutional review board, with signed informed consent obtained from the parents or guardians and assent from the patients, as appropriate. Long-term follow-up has been a major research emphasis in these trials. Survivors who have been in remission for 5 years or more are evaluated annually in our After Completion of Therapy Clinic. Through our Cancer Registry, questionnaire is mailed each year to the alumni survivors who are 18 years of age or older. Since 2007, alumni survivors have also been eligible for periodic cancer-related risk-based consultations as part of the St. Jude Lifetime Cohort Study.

Statistical Analysis

Event-free survival and survival from diagnosis or the completion of treatment in continuous complete remission were estimated by the method of Kaplan-Meier. The cumulative incidence functions of post-treatment failure due to a specific cause (relapse, development of a second neoplasm, or death in remission) were estimated by the method of Kalbfleisch and Prentice,[16] and compared using Gray’s test. Deaths in remission and second neoplasms were considered competing events in the estimation of cumulative incidence of relapse. The confidence intervals of failure probabilities were constructed using the normal approximation.[16] In a few cases where this method created a negative lower limit, a nonparametric method[17] was applied. The Fine and Gray model for competing risks and Cox regression model were used to identify independent risk factors for relapse and event-free survival, respectively. The proportion hazards assumption for each variable included in the model was checked using the Schoenfeld residuals. The Hochberg and Benjamini adaptive step-down Bonferroni method was used to adjust for multiple comparisons.[18] The actuarial risk of failures among survivors in study 15 was estimated with the Poisson model. The upper confidence bound of the expected number of failures in a given amount of person-year follow up was determined by using the exact upper confidence bound of the Poisson rate based on the observed number of failures and person-year follow-up. To provide a conservative estimate of the actuarial risk of relapse among 4-year post-treatment event-free survivors, a sensitivity analysis was performed on the unlikely assumption of up to 4 additional hypothetical relapses occurring during 2 additional years of follow-up. The database updated on August 23, 2013 was used for this analysis. The median follow-up time for patients remaining in continuous remission was 15.8 years (range, 1.3 to 29.4 years) in all studies and 8.2 years (range, 1.3 to 12.9 years) for those in study 15. At the time of analysis, 95% of the survivors had had a follow-up visit within 2 years; only 2.0% of the patients lacked a documented contact within the previous 5 years.

RESULTS

Of the 1527 patients enrolled in Total Therapy studies 11 to 15, 1291 (84.5%) were in continuous complete remission at the end of treatment including patients who underwent hematopoietic stem cell transplantation in first complete remission (Figure 1). As shown in Table 1, treatment outcome (event-free survival, survival and cumulative risk of any relapse) did not differ significantly between patients treated in studies 11 and 12, or among those treated in studies 13A, 13B, and 14, leading us to combine these cohorts into two groups for subsequent analyses. Patients treated in study 15 had a superior treatment outcome compared to those in studies 13A, 13B and 14, and therefore constituted a third comparison group. The 10-year event-free survival rate increased progressively over the three treatment periods (P<0.001): 65.2% (95% CI, 61.3% - 69.1%) vs. 74.8% (95% CI, 70.9% - 78.7%) vs. 85.1% (95% CI, 79.0% - 91.2%). The improved result in study 15 extended to NCI standard-risk B-ALL (p=0.006), NCI high-risk B-ALL (p<0.0001), and T-ALL (p<0.001) (data not shown). This improvement was also apparent in the analysis of overall survival (P<0.001): 76.5% (95% CI, 73.0% - 80.0%) vs. 81.1% (95% CI, 77.6% - 84.6%) vs. 91.7% (95% CI, 87.0% - 96.4%).

Figure 1

CONSORT diagram. Of the 562 patients treated in studies 11 and 12, 467 in study 13A, 13B and 14, and 498 in study 15, 438, 398, and 455 patients, respectively, remained in continuous complete remission upon the completion of treatment.

Table 1

Treatment Outcome According to Total Therapy Study

Total Therapy Studies	Total 11	Total 12	Total 13A	Total 13B	Total 14	Total 15
Year	1984-1988	1988-1991	1991-1994	1994-1998	1998-1999	2000-2007
No. of patients	374	188	167	247	53	498
Age (years)	0-18	0-18	0-18	0-18	0-18	1-18
Event-free survival % (95% CI)
5 years	70.0(65.3-74.7)	67.6(60.9-74.3)	76.7(70.2-83.2)	80.0(74.9-85.1)	77.4(66.2-88.6)	87.3(84.4-90.2)
10 years	67.3(62.6-72.0)	61.2(54.3-68.1)	70.7(63.8-77.6)	77.5(72.2-82.8)	75.5(64.1-86.9)	85.1(79.0-91.2)
Overall survival % (95% CI)
5 years	78.1(74.0-82.2)	83.5(78.2-88.8)	82.1(76.4-87.8)	86.1(81.8-90.4)	81.1(70.7-91.5)	93.5(91.3-95.7)
10 years	75.4(71.1-79.7)	78.7(72.8-84.6)	77.4(71.1-83.7)	84.1(79.4-88.8)	79.2(68.4-90.0)	91.7(87.0-96.4)
Off-therapy events
Hematological relapse	25	8	9	17	1	19
CNS relapse	3	3	0	2	0	1
Hematological + CNS relapse	3	5	4	2	1	5
Testicular relapse	0	2	0	1	0	2
Hematological + testicular relapse	1	2	1	0	0	0
Hematological + CNS + testicular relapse	1	0	0	0	0	0
Other relapse	1	1	0	1	0	0
Second malignancy	14	13	9	5	0	3
Death in remission	3	2	6	6	2	7
Off-therapy outcome % (95% CI)
5-yr event-free survival	86.7(82.8-90.6)	80.7(74.2-87.2)	86.4(80.7-92.1)	86.5(82.0-91.0)	90.5(76.6-96.3)	92.4(89.3-95.5)
10-yr event-free survival	85.3(81.2-89.4)	76.5(69.6-83.4)	82.1(75.8-88.4)	84.2(79.3-89.1)	90.5(76.6-96.3)	90.0(84.6-93.3)
5-yr survival	91.5(88.4-94.6)	91.0(86.3-95.7)	90.0(85.1-94.9)	91.6(87.9-95.3)	92.9(79.5-97.6)	95.9(93.5-98.3)
10-yr survival	90.4(87.1-93.7)	88.3(83.0-93.6)	87.9(82.4-93.4)	90.2(86.3-94.1)	92.9(79.5-97.6)	95.1(92.3-96.8)
5-yr cumulative risk of relapse	10.2(6.8-13.7)	13.1(7.6-18.6)	7.9(3.4-12.3)	9.3(5.4-13.2)	2.4(1.2-12.2)	5.9(3.7-8.1)
10-yr cumulative risk of relapse	11.3(7.6-14.9)	13.8(8.2-19.4)	10.0(5.0-15.0)	10.7(6.6-14.8)	4.8(1.2-12.2)	6.4(4.0-8.8)
5-yr cumulative risk of second malignancy	2.7(0.9-4.6)	5.5(1.8-9.2)	5.0(1.4-8.6)	2.3(0.3-4.3)	0	0.2(0.2-1.8)
10-yr cumulative risk of second malignancy	3.1(1.1-5.1)	9.0(4.3-13.7)	6.4(2.3-10.5)	2.3(0.3-4.3)	0	2.3(0.5-6.6)
5-yr cumulative risk of death in remission	0.3(0.05-1.4)	0.7(0.1-2.8)	0.7(0.4-3.9)	1.9(0.7-3.9)	4.8(1.2-12.3)	1.5(0.4-2.7)
10-yr cumulative risk of death in remission	0.3(0.50-1.4)	0.7(0.1-2.8)	1.4(0.4-3.9)	2.8(1.4-5.1)	4.8(1.2-12.3)	1.5(0.4-2.7)

CNS central nervous system

Major adverse events after completion of therapy

Of the 191 major post-therapy adverse events, hematological relapse in 104 cases (79 isolated and 25 combined with extramedullary relapse) was the most common, followed by the development of a second neoplasm in 44 cases (20 acute myeloid leukemias, 4 myelodysplastic syndromes, 1 chronic myeloid leukemia, 14 brain tumors and 5 solid tumors), and 17 extramedullary relapses (9 central-nervous-system, 5 testicular and 3 other sites). The remaining events comprised 26 deaths in remission due to various causes (10 accidents, 6 infections, 3 graft-versus-host disease, 3 multi-organ failure, 2 suicides, and 1 each of seizure and complications of ataxia-telangiectasia),12 of the deaths resulting from complications of hematopoietic stem cell transplantation. In study 15, 27 patients relapsed, 7 died in remission (3 graft-versus-host disease, 3 infections, and 1 accident), and 3 developed a second neoplasm (1 glioblastoma multiforme, 1 myelodysplastic syndrome, and 1 malignant fibrous histiocytoma) after completion of therapy among the 455 patients who completed all treatment in continuous complete remission.

Most of the adverse events (76 cases, 39.8%) occurred in the first year after completion of treatment, 40 (20.9%) in the second year, 13 (6.8%) in the third, 16 (8.4%) in the fourth, 7 (3.66%) in the fifth, 25 (13.09%) between the sixth and the tenth years, and 14 (7.33%) beyond 10 years (Supplementary Figure 1). Of the adverse events occurring after 10 years, 2 were leukemic relapses, 3 brain tumors, 2 solid tumors, and 7 deaths in remission.

Factors associated with the risk of a post-therapy adverse event

There was steady improvement in the proportion of patients who remained event-free after completion of therapy (p=0.004): 82.4 (95% CI, 78.9% - 85.9%) in studies 11 and 12 vs. 84.1% (95% CI, 80.6% - 87.6%) in studies 13A, 13B and 14 vs. 90.0% (95% CI, 84.6% - 93.3%) in study 15 at 10 years. DNA index ≥1.16, hyperdiploidy>50, absence of the t(9;22)/BCR-ABL1 or t(4;11)/MLL-AFF1, lack or low levels of minimal residual disease at day 19 of induction or the end of remission induction, and treatment according to study 15 were each associated with a higher post-therapy event-free survival rate (Supplementary Table 1). Treatment according to study 15, absence of t(9;22)/BCR-ABL1, hyperdiploidy>50, and lack of minimal residual disease at the end of remisson induction were independently associated with a favorable long-term outcome in a multivariate analysis (Supplementary Table 3).

In view of the decreasing use of carcinogenic treatment (i.e., epipodophyllotoxins and prophylactic cranial irradiation) in successive Total Therapy studies, we also compared the cumulative risk of a second neoplasm among the different groups (Figure 2A): 5.0% (95% CI, 3.0% to 7.1%) in studies 11 and 12 vs. 3.5% (95% CI, 1.7% to 5.5%) in studies 13A, 13B and 14 vs. 2.3% (95% CI, 0.5% to 6.6%) in study 15 at 10 years (p=0.003).

Figure 2

Cumulative risk of development of a second neoplasm (A) or relapse (B) after completion of therapy for patients in continuous remission according to Total Therapy study. The numbers of patients who remained event-free and were at risk of developing second neoplasm or relapse at any given year after completion of treatment for the three cohorts were provided in the bottom of the figure. Note the significant decrease in both adverse events for each successive treatment group.

The cumulative risk of any relapse following completion of therapy was 12.1% (95% CI, 9.0% to 15.2%) in studies 11 and 12, 9.8% (95% CI, 6.9% to 12.8%) in studies 13A, 13B and14, and 6.4% (95% CI, 4.0% to 8.8%) in study 15 at 10 years (p=0.026; Figure 2B). A lower risk of post-therapy relapse was associated with treatment according to study 15, female sex and the absence of t(9;22)/BCR-ABL1 in both univariate (Table 2) and multivariate (Table 3) analyses. Among the subgroup of 629 patients with data on minimal residual disease, only the lack (<0.01%) and low level (0.01% to <0.1%) of minimal residual disease at the end of remission induction therapy were associated with lower risk of post-therapy relapse (Table 3).

Table 2

Factors associated with cumulative risk of any post-therapy relapse and time to relapse

Factor	No. Patients	No. Patients withPost-TherapyRelapse	Median time to Post-Therapy Relapse(range) in Years	Estimate % (95% CI)
				Year 5	Year 10	P Value
NCI risk group (B- ALL)
Standard	671	55	1.36(0.07-9.23)	7.6(5.6-9.6)	8.6(6.4-10.8)	1.00
High	418	43	1.15(0.00-9.14)	8.9(6.2-11.7)	10.8(7.7-13.9)
NCI risk group (T- ALL)
Standard	25	3	0.73(0.52-20.51)	8.0(0.0-18.9)	8.0(0.0-18.9)	1.00
High	145	14	1.17(0.15-10.15)	8.4(3.8-12.9)	9.2(4.4-14.1)
Sex
Male	688	81	1.15(0.07-20.5)	10.4(8.1-12.7)	12.2(9.6-14.7)	0.02
Female	603	40	1.34(0.00-7.10)	6.2(4.2-8.1)	6.8(4.8-8.8)
Age at Diagnosis
Infant	20	2	0.72(0.54-0.91)	10.0(0.0-23.6)	10.0(0.0-23.6)	1.00
1-10 years	1005	94	1.26(0.07-20.51)	8.4(6.6-10.1)	9.6(7.7-11.4)
Older than 10	266	25	1.32(0.00-7.48)	8.5(5.1-11.9)	10.0(6.2-13.7)
Leukocyte count at Diagnosis
<10 × 109/L	604	54	1.30(0.13-9.23)	8.2(6.0-10.4)	9.3(6.9-11.7)	1.00
10 to 49 × 109/L	387	38	1.81(0.07-20.51)	8.7(5.9-11.6)	10.1(7.0-13.2)
50 to 99 × 109/L	142	11	1.12(0.44-5.60)	6.5(2.4-10.6)	8.2(3.5-12.9)
≥100 × 109/L	158	18	0.55(0.00-10.15)	10.2(5.4-14.9)	11.0(6.0-16.0)
Race
White	1059	93	1.32(0.00-20.51)	7.9(6.2-9.5)	9.0(7.2-10.8)	1.00
Black	180	20	0.82(0.16-9.14)	10.1(5.7-14.6)	11.8(6.8-16.7)
Other	52	8	1.47(0.44-6.09)	13.5(4.1-22.8)	15.8(5.6-26.1)
Immunophenotype
B	1089	98	1.26(0.00-9.23)	8.1(6.5-9.8)	9.5(7.7-11.3)	1.00
T	170	17	1.02(0.15-20.51)	8.3(4.1-12.5)	9.0(4.7-13.4)
CNS status
CNS-1	913	85	1.32(0.13-20.51)	8.4(6.5-10.2)	9.7(7.7-11.6)	1.00
CNS-2	251	26	1.40(0.00-10.15)	8.8(5.3-12.4)	10.5(6.5-14.4)
Traumatic lumber puncture with blast	89	5	1.50(0.11-3.56)	5.8(0.8-10.8)	5.8(0.8-10.8)
CNS-3	33	4	0.78(0.27-1.15)	12.1(0.8-23.5)	12.1(0.8-23.5)
DNA index
≥1.16	295	20	1.64(0.31-4.59)	6.9(4.0-9.9)	6.9(4.0-9.9)	0.95
<1.16	996	101	1.15(0.00-20.51)	8.9(7.1-10.6)	10.4(8.5-12.4)
Genetic abnormality
Hyperdiploidy >50
Present	381	30	1.39(0.07-6.37)	7.5(4.8-10.1)	8.1(5.3-10.9)	1.00
Absent	856	82	1.02(0.00-20.51)	8.3(6.4-10.2)	9.8(7.8-11.9)
t(9;22)/BCR-ABL1
Present	31	9	0.43(0.11-1.84)	29.0(12.7-45.4)	29.0(12.7-45.4)	0.002
Absent	1260	112	1.52(0.12-20.51)	7.9(6.4-9.4)	9.2(7.5-10.8)
t(1;19)/TCF3-PBX1
Present	60	1	0.97(0.97-0.97)	1.7(0.0-4.9)	1.7(0.0-4.9)	0.52
Absent	1231	120	1.26(0.00-20.51)	8.7(7.1-10.3)	10.0(8.3-11.7)
t(12;21)/ETV6-RUNX1
Present	223	13	2.89(0.46-9.23)	4.1(1.5-6.8)	6.8(3.1-10.5)	0.55
Absent	1068	108	1.13(0.00-20.51)	9.3(7.6-11.1)	10.3(8.4-12.1)
t(4;11)/MLL-AFF1
Present	15	2	1.08(0.91-1.25)	13.3(0.02-31.3)	13.3(0.03-31.3)	1.00
Absent	1276	119	1.28(0.00-20.51)	8.4(6.8-9.9)	9.6(7.9-11.3)
Minimal residual disease
On day 19 of induction
<0.01%	255	15	1.80(0.12-5.73)	5.7(2.8-8.6)	6.4(3.2-9.5)	1.00
0.01%-<0.1%	126	7	1.99(0.13-9.14)	4.8(1.0-8.6)	8.9(0.1-17.7)
≥0.1%	194	20	1.11(0.00-6.09)	10.1(5.8-14.4)	11.1(6.4-15.7)
On remission date
<0.01%	507	27	1.80(0.12-7.48)	4.9(3.0-6.8)	6.0(3.7-8.3)	<0.001
0.01%-<0.1%	63	4	2.53(0.93-9.14)	5.0(0.03-10.6)	9.8(0.03-20.7)
≥0.1%	59	15	1.08(0.00-3.63)	25.9(14.4-37.3)	25.9(14.4-37.3)
Total Therapy Studies
11 and 12 (1984-1991)	438	55	1.15(0.15-20.51)	11.2(8.2-14.1)	12.1(9.0-15.2)	0.026
13A, 13B and 14 (1991-1999)	398	39	1.78(0.11-9.23)	8.1(5.4-10.7)	9.8(6.9-12.8)
15 (2000-2007)	455	27	1.03(0.00-6.09)	5.9(3.7-8.1)	6.4(4.0-8.8)

Table 3

Independent risk factors for post-therapy relapse

Features	Analysis excluding MRD		Analysis including MRD
	Hazard ratios (95% CI)	P value	Hazard ratios (95% CI)	P value
Study 15	0.61(0.4-0.96)	0.03	0.66 (0.35-1.23)	0.18
BCR-ABL1 Absent	0.27 (0.12-0.57)	<0.001	0.44 (0.11-1.64)	0.22
Negative (<0.01%) MRD onremission date	---------	-------	0.26(0.11-0.54)	<0.001
0.01%-<0.1% MRD on remissiondate	---------	-------	0.27(0.08-0.87)	0.028
Female	0.58(0.4-0.88)	0.01	0.53(0.27-1.04)	0.06

MRD denotes minimal residual disease, CI confidence interval

Time to relapse

The time to relapse varied widely according to the clinical or biologic factors examined (Table 2). Notably, no relapses beyond 2 years were observed in the 106 patients with the t(9;22)/BCR-ABL1, t(1;19)/TCF3-PBX1 or t(4;11)/MLL-AFF1 (Table 2).

Analysis of the conditional probability of post-therapy relapse for each treatment group (Figure 3) showed that there was still an estimated risk of relapse of 0.65% (95% CI, 0.04% to 1.17%) beyond 10 years after completion of therapy in studies 11 and 12, whereas no relapse was observed beyond 10 years in studies 13A, 13B and 14. In study 15, there were 13 relapses in the first year after completion of treatment, 7 between 1 and 2 years, 4 between 2 and 3 years, 2 between 3 and 4 years, and 1 at 6 years. Of the 418 patients in study 15 who completed all treatment and remained alive and event-free at the time of analysis, 313 (75%) had been followed for 4 years or more after completion of therapy. With 827.6 person-years of follow-up (after completion of therapy for 4 years) of the 313 patients, the 95% and 99% upper confidence bound of the expected number of relapse per 100 person-years of follow-up beyond 4 years after completion of treatment are 0.57 and 0.8, respectively. To provide a conservative estimate of the future risk of relapse among the 4-year post-treatment event-free survivors, we performed a sensitivity analysis covering a range of likely to unlikely scenarios, in which 1 to 4 hypothetical additional relapses will occur during 2 additional years of follow up. The analyses showed the 99% upper confidence bound of the actuarial risk of 0.42 to 0.74 relapse per 100 person-years of follow-up (Supplementary Table 4), indicating an extremely low risk of further relapse.

Figure 3

Conditional probability of relapse for patients remaining in continuous complete remission for the given years after completion of treatment by Total Therapy study. Note the progressive decrease in the probability of off-therapy relapse for each successive treatment group, and the absence of relapse in patients remaining in remission at 4 years off therapy in study 15.

DISCUSSION

This study demonstrates that the improvement in treatment of childhood ALL since the mid-1980s has significantly reduced not only the risk of leukemic relapse but also the risk of developing a second neoplasm. Indeed, the cumulative risk of any relapse at 10 years after completion of therapy was reduced from 12.1% in studies 11 and 12 to 9.8% in studies 13A, 13B and 14 to only 6.4% in study 15. Equally important, the duration of risk of relapse after completion of treatment also decreased progressively, with only one of the 313 patients in study 15 who had been off treatment and event-free for 4 years relapsing beyond this time point. Even with additional follow-up, the actuarial risk of relapse should not exceed 1 per 100 person-year follow up for the 4-year post-treatment event-free survivors in this study.

Undoubtedly, the complete omission of prophylactic cranial irradiation and the restriction of administration of epipodophyllotoxins to patients undergoing transplantation were crucial in reducing the risk of a second neoplasm to a very low level in study 15. Among 498 patients treated in that study, we observed only one case each of myelodysplastic syndrome,[7] primary malignant fibrous histiocytoma and glioblastoma multiforme after cessation of treatment. Although antimetabolite treatment has been associated with the development of secondary myeloid neoplasm, especially in patients with thiopurine methyltransferase deficiency,[19] the child who developed myelodysplastic syndrome had normal thiopurine methyltransferase activity. The patient who developed glioblastoma multiforme had received total-body irradiation as part of the preparative regimen for transplantation. In this regard, prophylactic cranial irradiation as low as 12 Gy can be associated with the development of brain tumor.[20] Whether the patient who developed a malignant fibrous histiocytoma has underlying genetic susceptibility to develop cancer is unknown.

Late relapse of ALL has been variously attributed to the outgrowth of a minor drug-resistant subclone, acquisition of additional genetic abnormalities by a resilient preleukemic clone, or the de novo development of secondary ALL.[21-27] Conceivably, early intensification with asparaginase and dexamethasone as well as high-dose methotrexate, which were applied to all patients and at higher doses for those with high-risk B-ALL and T-ALL, likely eliminated a higher proportion of mutant stem-cell lines in study 15 than did therapies administered in the five preceding trials (see Supplementary Table 2). The restriction of epipodophyllotoxins might also have prevented de novo development of secondary ALL in this study.

The presence of the t(9;22)/BCR-ABL1 was one of the factors significantly associated with the post-treatment relapse hazard. However, ABL tyrosine kinase inhibitors, effective agents for t(9;22)/BCR-ABL1-positive ALL,[28,29] were available for only the last few patients enrolled in study 15. The presence of minimal residual disease at the end of remission induction is one of the strongest prognostic factors in childhood ALL and correlated with a high cumulative risk of relapse in virtually all clinical trials.[30-35] In a Children’s Oncology Group study, detectable minimal residual disease at the end of remission induction was also associated with a greater risk of relapse at 3 years or more after diagnosis,[32] a finding confirmed in our study.

In the Children’s Oncology Group study, the presence of favorable genetic features such as ETV6-RUNX1 or trisomies of chromosomes 4 and 10 provided additional prognostic information beyond the absence of minimal residual disease at the end of remission induction.[32] However, neither ETV6-RUNX1 nor hyperdiploidy >50 (which is associated with trisomies of chromosomes 4 and 10) was independently related to the risk of post-treatment relapse in the current analysis. As reported previously,[36] gender remained a risk factor in this study.

The biologic subtypes of ALL can have a major influence on the time to relapse. Late relapse is well recognized in ETV6-RUNX1-positive and T-cell ALL,[25,37-39] and has been attributed to a second, independent malignant transformation of a preleukemic clone that was not eradicated during initial treatment. While we also observed late relapses in 4 patients with ETV6-RUNX1-positive ALL treated in studies 12, 13A and 13B at 5.1, 5.6, 9.1 and 9.2 years after completion of treatment, none of the 59 patients with ETV6-RUNX1-positive ALL treated in study 15 relapsed beyond 4 years after completion of treatment, suggesting that more effective treatment can eradicate preleukemic clones in such patients. Finally, relapses were not observed beyond 2 years after completion of therapy among 106 patients with the t(9;22)/BCR-ABL1, t(1;19)/TCF3-PBX1 or t(4;11)/MLL-AFF1 in any of our Total Therapy studies, indicating that these patients may be declared cured earlier than patients with other biologic subtypes of ALL.

Our results allow us to revise the timing for an objective definition of cure. We propose that, with use of effective treatment similar to that in study 15, a patient who did not receive cranial irradiation and remains in complete remission for at least 4 years after the completion of antileukemic therapy should practically be considered cured because the likelihood of subsequent late event is exceedingly rare. This new definition should be applicable to contemporary clinical trials that shared similar treatment approach with intensified dexamethasone and asparaginase and achieved 5-year event-free survival rates of 85% or higher and 5-year survival rates of 90% or more.[6-10] It should also alleviate much of the uncertainty and stress surrounding elective cessation of therapy following prolonged continuation treatment,[40] and help to overcome barriers to insurance coverage and health-care access experienced by many long-term survivors.[1,41,42] Finally, the omission of cranial irradiation will contribute to preserving global cognitive abilities, including intellectual functioning, academic abilities, learning and memory, as reported for patients treated in study 15.[43]