Frequencies of SF3B1, NOTCH1, MYD88, BIRC3 and IGHV mutations and TP53 disruptions in Chinese with chronic lymphocytic leukemia: disparities with Europeans.

We studied 307 consecutive Chinese with chronic lymphocytic leukemia (CLL) in diverse disease-stages before and after diverse therapies for mutations in several CLL-related genes. Mutation frequencies were SF3B1, 5%, NOTCH1, 8%, MYD88, 8%, BIRC3, 2%, TP53, 15% and IGHV, 60%. Several of these frequencies differ from those reported in persons of predominately European descent with CLL. Biological and clinical associations were detected including SF3B1 and NOTCH1 mutations with un-mutated IGHV, MYD88 mutations with mutated IGHV, SF3B1 mutations with fludarabine-resistant CLL and NOTCH1 mutation with advanced Binet disease stage and with +12. The NOTCH1 correlation with briefer survival was confirmed in multivariate analyses but the SF3B1 correlation was confounded by concurrent mutations in TP53 and germline IGHV. We show differences in incidence and prognostic impact of mutations in Chinese and CLL compared with persons of predominately European descent with CLL. These data may give insights into the etiology and biology of CLL and suggests different risk stratification models may be needed for different CLL populations.

INTRODUCTION

Biology and clinical course of chronic lymphocytic leukemia (CLL) are heterogeneous. An unfavorable prognosis is associated with certain cytogenetic and molecular abnormalities such as del(11q22-q23), TP53 disruption and un-mutated immunoglobulin heavy-chain variable region (IGHV) state [1-3]. Mutations in RNA-splicing and processing, genes involved in Notch signaling, inflammatory pathway genes and genes in the Wnt pathway are also associated with briefer survival of persons with CLL [4].

Mutations in SF3B1, NOTCH1, MYD88 and BIRC3 are relatively rare in persons of predominately European descent with CLL with frequencies of 5-18%, 8-12%, ~3%, and ~2% [4-10]. However, mutations of these genes in a sub-clone of CLL cells undetectable by conventional sequencing may become detectable as CLL progresses. It is also likely some new mutations are acquired as CLL progresses with or without therapy [11]. We studied frequencies and prognostic associations of mutations of SF3B1, NOTCH1, MYD88, BIRC3, TP53 and IGHV and cytogenetic abnormalities in 307 Chinese with CLL.

RESULTS

Subjects

Characteristics of the 307 subjects are summarized in Table 1. Median age was 61 years (range, 16- 92 years) with a male/female ratio of 2:1. Median follow-up is 35 months (range, 1-267 months),

Table 1

Clinical characteristics

	N (%)
Sex (n=307)
Male	204 (66)
Female	103 (34)
Age (n=307)
≥60	164 (53)
<60	143 (47)
Binet (n=297)
A	129 (43)
B	70 (24)
C	98 (33)
Disease States (n=307)
At diagnosis	166 (54)
Progressive	88 (29)
Relapsed	25 (8)
Refractory	28 (9)

SF3B1, NOTCH1, MYD88, BIRC3 and TP53 mutations

307 subjects were studied for SF3B1 mutations. 15 (5%) had a SF3B1 mutation. All mutations were missense with p.K700E the most frequent (N=10). 24 of 295 (8%) of subjects analyzed had NOTCH1 mutations of whom 17 had c.7544-7545delCT. Other mutations were nonsense (N=3) or out-of-frame (N=4). Two mutations, p.L2049fs*1 and p.Q2404X, were not reported by COSMIC or other studies. Mutations in exons 3-5 of MYD88 were detected in 23 (8%) of 295 subjects analyzed. The most frequent mutation was p.L265P (N=15). p.V271F is a new mutation. 5 of 238 (2%) had a BIRC3 mutation including 4 out-of-frame and 1 in-frame. 47 (15%) of 307 subjects analyzed had TP53 mutations. Most mutations were in the DNA-binding domain (N=45; 94%).

Subjects with SF3B1 mutation were more likely to have concomitant TP53 mutation (5 of 15 vs. 42 of 292; P=0.047). Also, MYD88 mutations were exclusive of NOTCH1 and SF3B1 mutations. Detailed data are presented in Supplement Table 1.

IGHV Mutations

In 118 of 299 (40%) subjects IGHV was germline. The most frequently used IGHV genes were IGHV4-34 (N=38; 13%), IGHV3-23 (N=33; 11%), IGHV3-7 (N=30, 10%), IGHV4-39 (N=21; 7%), IGHV4-59 N=15; 5%), IGHV1-69 (n=15; 5%) and IGHV3-21 (N=9; 3%).

Clinical correlations

NOTCH1 mutations were detected more frequently in subjects with advanced Binet stages (stage-A, 4 of 126; stage-B: 6 of 66; stage-C, 14 of 94; P=0.008). TP53 mutations were also more common in subjects with advanced Binet stage (stage-A, 11 of 129; stage-B,12 of 70; stage-C, 23 of 98; P=0.008). We found no significant correlation between Binet stage and SF3B1 or MYD88 mutations (Table 3) but power to detect an association is limited because of the low frequency of mutations. MYD88 mutations were less frequent in subjects in whom CD38 was ≥30% (1 of 67 vs, 22 of 223; P=0.026). NOTCH1 mutations were associated with CD38 ≥30% but this difference was not significant (9 of 67 vs. 15 of 223; P=0.081). We found no correlation between SF3B1 or TP53 mutations and CD38 ≥30%. There was also no correlation between any of the mutations we studied and ZAP70 ≥20%. BIRC3 mutations were excluded from this analysis because they were so rare (Table 3).

Table 3

Associations between subject variables and mutations

		SF3B1mut	P	NOTCH1mut	P	MYD88mut	P	TP53mut	P
N	307	15/307 (5%)	-	24/295 (8%)	-	19/229 (8%)	-	47/307 (15%)	-
Female	103 (34%)	2/103 (2%)	0.089	7/101 (7%)	0.585	6/101 (6%)	0.391	21/103 (20%)	0.093
Male	204 (66%)	13/204 (6%)		17/194 (9%)		17/194 (9%)		26/204 (13%)
Age	61	61(±12)	0.911	64(±11)	0.212	62(±8)	0.631	61(±13)	0.890
Binet A	129(43%)	5/129 (4%)	0.305	4/126 (3%)	0.008	9/126 (7%)	0.680	11/129 (9%)	0.008
Binet B	70 (24%)	6/70 (9%)		6/66 (9%)		4/66 (6%)		12/70 (17%)
Binet C	98 (33%)	4/98 (4%)		14/94 (15%)		9/94 (10%)		23/98 (24%)
CD38≥30%	71/302 (24%)	5/71 (7%)	0.356	9/67 (13%)	0.081	1/67 (2%)	0.026	10/71 (14%)	0.677
CD38<30%	231/302 (76%)	10/231 (4%)		15/223 (7%)		22/223 (10%)		36/231 (16%)
ZAP70≥20%	110/272 (40%)	7/110 (6%)	0.613	11/106 (10%)	0.246	6/106 (6%)	0.523	17/110 (16%)	0.774
ZAP70<20%	162/272 (60%)	8/162 (5%)		10/156 (6%)		12/156 (8%)		23/162 (14%)
IGHV M	181/299 (61%)	5/181 (3%)	0.027	2/172 (1%,)	<0.001	21/172 (12%)	0.001	19/181 (11%)	0.001
IGHV UM	118/299 (40%)	10/118 (9%)		21/115 (18%)		2/115 (2%)		28/118 (24%)

SF3B1, NOTCH1 and TP53 mutations were significantly correlated with germline IGHV (SF3B1, 10 of 118 vs. 5 of 181; P=0.027; NOTCH1, 21 of 115 vs. 2 of 170; P<0.001; TP53, 28 of 118 vs. 19 of 181: P=0.002). In contrast, MYD88 mutations were more common in subjects with mutated IGHV (2 of 115 vs. 21 of 172; P=0.001; Table 3). NOTCH1 mutations were especially common in subjects using IGHV4-39 (4 of 19 vs. 19 of 268; P=0.03) and IGHV1-69 (6 of 15 vs. 17 of 255; P<0.001). SF3B1 mutations were correlated with IGHV4-59 (4 of 15 vs. 11 of 284; P=0.004).

Association with cytogenetic abnormalities

Cytogenetic abnormalities were common in our subjects including +12 in 59 of 257 (23%), del(17p13) in 42 of 281 (15%) and del(11q22.3) in 40 of 283 (14%). As expected, NOTCH1 mutations were significantly associated with +12 (9 of 56 vs. 12 of 192; P=0.020) and TP53 mutations with del(17p13) (23 of 40 vs. 22 of 243; P<0.001). Among the 64 subjects with TP53 abnormalities, 17 had only del(17p13), 23 had TP53 mutation and del(17p13) and 22 only TP53 mutation (2 missed del(17p13) data, but got TP53 mutation). Surprisingly, there was no correlation between SF3B1 mutations and del(11q22.3) (4 of42 vs. 11 of 239; P=0.191). MYD88 mutations did not correlate with cytogenetic abnormalities (Supplement Figure 1). Detailed distribution of mutations and cytogenetic lesions are shown in Figure 1.

Figure 1

Relationship between TP53, SF3B1, NOTCH1 and MYD88 mutations, and cytogenetic abnormalities in 307 subjects with data

Cases which presented with a cytogenetic aberration or a mutation are colored in red. Cohort1: Newly-diagnosed; cohort 2: Progressive; no therapy; cohort 3:Relapsed CLL after therapy; and cohort 4: Refractory, therapy-resistant. Subjects are highlighted in light blue, pink, light green and light purple, respectively. Brown means not analyzed.

Prognostic relevance of mutations and cytogenetic abnormalities

We divided subjects into 4 prognostic cohorts: (1) newly-diagnosed; (2) progressive; no therapy; (3) relapsed CLL after therapy; and (4) refractory, fludarabine-resistant. Frequency of gene mutations in each cohort is shown in Table 2. Mutation rate of SF3B1 was significantly higher in subjects in cohort 4 (P=0.007). NOTCH1 was less frequently mutated in subjects in cohort 1 (P<0.001) but similar in cohorts 2-4 (P>0.05). TP53 mutation was less frequent in subjects in cohort 1 (P=0.004) and more common in subjects in cohort 4 (P=0.033). MYD88 mutation was more common in subjects in cohort 1 and 2 (P<0.001). There was no significant association of BIRC3 mutation with any cohort.

Table 2

Mutation frequency by disease state

	Diagnosis	Progressive	Relapsed	Refractory
SF3B1	6/166 (4%)	4/88 (5%)	0/25	5/28 (18%)
NOTCH1	4/158 (3%)	9/84 (11%)	4/25 (16%)	7/28 (25%)
MYD88	15/158 (10%)	7/84 (8%)	1/25 (4%)	0/28
BIRC3	2/119 (2%)	1/71 (1.4%)	2/21 (10%)	0/27
TP53	13/166 (8%)	18/84 (21%)	4/25 (16%)	12/28 (43%)

Prognostic impact of the abnormalities we studied is shown in Figure 2. Survival was significantly briefer in subjects with NOTCH1 mutation (median, 63 vs. 153 months; P=0.008), del(11q22.3) (median, 77 vs. 152 months; P=0.030) and TP53 disruptions (median, 71 months vs. NR, P<0.001). MYD88 mutations had no significant impact on survival (median, NR vs. 142 months; P=0.590). However, SF3B1 mutations had only a borderline impact on survival (median, 70 vs. 152 months; P=0.055) even when 6 subjects with concurrent TP53 mutations were included. The frequency of BIRC3 mutations in BIRC3 was too low for statistical comparisons (Supplement Table 2 ).

Figure 2

Kaplan-Meier curves of survival (a-c) and time-to-treatment (TTT) (d-f) for SF3B1 (a and d), NOTCH1 (b and e) and MYD88 mutations (c and f)

We divided the TP53 mutation cohort into 2 cohorts based on IGHV rearrangement. 30 of 64 subjects with TP53 disruption and mutated IGHV had longer survival compared with those with TP53 disruption and germline IGHV (median, 141 vs. 60 months; P=0.001). Survival of the favorable cohort was not significantly different from subjects without TP53 disruptions (median, 141 months vs. NR; P=0.308; Figure 3a). There was also no significant survival difference with only TP53 mutation, only del(17p13) or both (P=0.474).

Figure 3

Kaplan-Meier curves of survival (a) and TTT (b) for TP53 disruption cases with different IGHV genemutation states

Comparison of survival subjects with multiple unfavorable alterations including TP53 vs. those with TP53 only mutation is shown in (c).

Next, we compared whether subjects with TP53 disruptions and other unfavorable abnormalities including del(11q22.3) or mutations in NOTCH1 and/or SF3B1 mutations had shorter survival compared with those with TP53 disruptions only. There was no significant difference (median, 70 vs. 71 months; P=0.540; Figure 3c). Multiple unfavorable abnormalities were more common in subjects with refractory, fludarabine-resistant disease (cohort 4; 6 of 143 vs. 14 of 127; P=0.033; Supplement Table 3).

The impact of each mutation on TTT was studied in previously untreated subjects (cohort 1 and 2; Table 4). Subjects with NOTCH1 mutations and/or TP53 disruptions had briefer TTT than subjects without these abnormalities (median, 2 vs. 57 months; P<0.001 and 12 vs. 62 months; P=0.003). Subjects with TP53 disruptions and mutated IGHV had similar TTT as subjects without TP53 disruptions (median: 46 vs. 62 months; P=0.423). There was no significant correlation between mutations in SF3B1 or MYD88 mutation and TTT (median, 14 vs. 46 months; P=0.586; median, 19 vs. 46 months; P=0.619; Figure 2 and 3b).

Table 4

Associations between mutations, cytogenetics, time-to-treatment and survival

	Time-to-treatment			Survival
Variable	N	Median (mo)	P-value	N	Median (mo)	P-value
TP53 disruption	234		0.003	285		<0.001
Yes	44	12		64	71
No	190	62		221	NR
del(11q22.3)	232		0.061	281		0.030
Yes	28	18		42	77
No	204	47		239	152
+12	211		0.393	257		0.366
Yes	44	37		59	117
No	167	66		198	152
SF3B1	254		0.586	307		0.055
Mutated	10	14		15	72
Wild-type	244	46		292	152
NOTCH1	242		<0.001	295		0.008
Mutated	13	2		24	63
Wild-type	229	57		271	152
MYD88	242		0.619	295		0.589
Mutated	22	19		23	NR
Wild-type	220	46		272	141

In multivariate analyses variables independently-correlated with TTT included germline IGHV (HR=2.30, 95% CI: 1.51–3.49; P<0.001), TP53 disruptions (HR=1.85, 95% CI: 1.20–2.83; P=0.005) and NOTCH1 mutation (HR=2.17, 95% CI: 1.11–4.23; P=0.024).

DISCUSSION

We provide data of frequencies of mutations in SF3B1, NOTCH1, MYD88, BIRC3 in Chinese with CLL along with data on previously described prognostic variables including TP53 disruptions, IGHV mutation and cytogenetic abnormalities. One striking difference was the frequency, biological features and prognostic impact of SF3B1 mutations in Chinese with CLL. The frequency we detected is considerably lower than reported in persons of predominately European descent (Table 2) [4, 7, 8, 12-14]. This low rate might be accounted for by the relatively low sensitivity of the Sanger sequencing we used. For example, Jeromin et al. reported about 10% SF3B1 mutations occurred at a mutation load ≤10%, a level detectable only with next generation sequencing (NGS) and would have been missed by us [13]. However, this is unlikely to explain the disparate frequencies we observed. We found SF3B1 mutation frequency was significantly higher in subjects with advanced, fludarabine-resistant CLL consistent with the notion detection of these mutations result from expansion of a sub-clone of CLL cells and are acquired during disease progression [15, 16].

We also found discrepancies regarding to IGHV gene use by SF3B1 mutated cases between our subjects and persons of predominately European descent. Strefford et al. reported a much higher frequency of SF3B1 mutations in stereotyped IGHV3-21 [17]. A recent study of 1160 untreated persons of predominately European descent with CLL reported frequent SF3B1 mutations in subjects with IGHV3-21 and IGHV1-69 gene use [13]. Many of our subjects had germline IGHV. In contrast, SF3B1 mutations were not correlated with IGHV1-69 or IGHV3-21 but rather with IGHV4-59. This low frequency of IGHV1-69 use in Chinese with CLL is previously reported [18, 19]. Only 9 of 299 subjects had IGHV3-21 use and only 15 used IGHV1-69. Furthermore, only one IGHV1-69 user had SF3B1 p.K700E mutation. The reason for the disparities between Chinese and persons of European descent with CLL are unknown and could relate to the low incidence of SF3B1 mutations and IGHV1-69 and IGHV3-21 use in Chinese with CLL. It is also possible different antigenic stimuli operate in diverse geographies [20]. Interestingly, SF3B1 mutation was not associated with advanced Binet stage, CD38 ≥30% or del(11q22.3) but was correlated with TP53 mutation. These data suggest distinct biological features of SF3B1 mutated cases of CLL in Chinese may differ from those in persons of predominately European decent.

Incidence of NOTCH1 mutations in our cohort is relatively low at diagnosis but increased with disease progression. These data are compatible with a sub-clone of CLL cells below the sensitivity of Sanger sequencing followed by clonal expansion during disease course, acquisition of additional mutations spontaneously for as a consequence of therapy or both [11, 16]. Mutation detection methods with higher sensitivity, such as allele specific PCR, should be able to distinguish these possibilities which are not mutually-exclusive [21]. The hot spot c.7544-7545delCT deletion accounted for about 70% of our mutated cases. Notably, two novel nonsense mutations were found suggesting the possibility of novel NOTCH1 mutation sites. Significant associations were observed between NOTCH1 mutations and advanced Binet stage, germline IGHV and +12. IGHV4-39 and IGHV1-69 were frequently used in subjects with NOTCH1 mutations. +12 and NOTCH1mutations are reported to characterize IGHV4-39 CLL belonging to subset 8 and with higher risk of transformation to diffuse large B-cell lymphoma (DLBCL; Richter syndrome) [22, 23]. We observed only 1 case of transformation, possibly because of brief follow-up.

Another interesting finding is the high frequency of MYD88 mutation in our cohort [12-14]. MYD88 mutation is common in subjects with mutated IGHV and rare in the CD38 ≥30% cohort. Moreover, almost all MYD88 mutations occurred in untreated subjects consistent with the hypothesis MYD88 mutation occurs early in CLL development [11]. In addition, we found no significant correlation between BIRC3 mutation in fludarabine-resistant patients as previously reported [10].

We show NOTCH1 and TP53 disruptions are unfavorable factors for survival and TTT consistent with prior studies in persons of predominately European descent [5, 6, 8, 11-14, 16]. In contrast, the prognostic impact of SF3B1 mutations is modest, if any. The unfavorable impact of SF3B1 mutation on survival was largely dependent on concurrent mutations in TP53 and germline IGHV. Our data contradict other studies in persons of predominately European descent in whom SF3B1 mutation is a strong independent predictor of brief survival [4, 7, 8, 10, 11-16]. We had, of course, little power to test the impact of SF3B1 mutation alone on survival because of the low frequency so our conclusion should be viewed cautiously.

We found high rate of subjects with TP53 mutations in persons without del(17p13). These subjects had a poor prognosis similar to those with both abnormalities similar to our prior report [24]. However, we also found a cohort of subjects with TP53 disruptions had stable disease. Subjects with TP53 disruptions and germline IGHV had the worst prognosis whereas those with TP53 disruptions and mutated IGHV gene had longer TTT and survival similar to subjects without TP53 abnormality. Others report similar data emphasizing the combination of IGHV rearrangement and TP53 disruptions might be a better predictor of prognosis than either variable alone [14, 25].

Recently, Greipp et al. reported persons with CLL with del(17p13) and del(11q22.3) had briefer survival compared with persons with del(17p13). These persons were termed double hit CLL [26]. Based on this report we compared outcomes of persons with TP53 disruptions and other unfavorable abnormalities including del(11q22.3) or mutations in NOTCH1 and/or SF3B1 mutations with those with TP53 mutation only. We found no significant difference in outcomes. There are several possible explanations for this disparity including different populations, different NOTCH1 and SF3B1 mutations and different cut-off values for FISH in the 2 studies. Nevertheless, our data support the concept many unfavorable genetic abnormalities are more common in advanced CLL.

In multivariate analyses, NOTCH1 mutation, germline IGHV and TP53 disruptions were independently correlated with TTT. Detection of NOTCH1 mutations helped us reclassify 13 subjects with no other unfavorable risk factors. Patients with NOTCH1 and/or SF3B1 mutations were considered intermediate to high risk in other reports [12-14]. In our cohort NOTCH1 mutation had the greatest risk of requiring therapy which would also indicate high-risk disease. In contrast, SF3B1 mutations had little or no prognostic impact and were included in the good-risk cohort. These disparities highlight the need for different risk classifications for different populations with CLL.

In conclusion, we studied frequency and prognostic impact of cytogenetic and molecular abnormalities in a large series of Chinese with CLL and compared these data with data from persons of predominately European descent with CLL. Our study highlights important similarities but also which may offer important clues to the etiology and biology of CLL.

METHODS

The study cohort was a single-center consecutive series of 307 Chinese with CLL. Subjects were diagnosed from November, 1991 to April, 2014. Subjects provided informed consent according to institutional guidelines. Diagnosis of CLL was based on International Workshop on CLL-National Cancer Institute (IWCLL-NCI) criteria [27].

Analyses of SF3B1, NOTCH1, MYD88 mutations and TP53 disruptions

Genomic DNA was isolated from mononuclear cells using the QIAamp DNA Blood Kits (Qiagen, Düsseldorf, Germany) according to the manufacturer's recommendation. Direct Sanger sequencing was performed for exon 14-16 of SF3B1, PEST domain of NOTCH1, exon 3-5 of MYD88, exon 6-9 of BIRC3 and exon 4-9 of TP53. Primers are listed in Supplement Table 4.

Cytogenetics

Fluorescence in situ hybridization (FISH) analysis was performed on most subjects to detect of del(11q22.3) (n=281), del(17p13) (n=283) and trisomy 12 (n=257). The following fluorescent-labeled probes were used: LSI ATM (11q22.3), LSI p53 (17p13) and CEP12 (centromere 12). Probes were purchased from Vysis, Downers Grove, IL, USA. FISH was performed as described [24]. Cut-off levels for positivity were 7.7%, 5.2%, and 3.0% for del(11q22.3), del(17p13) and +12. Although not all subjects with TP53 mutation were analyzed for del(17p13), we refer the cohort with TP53 mutation or del(17p13) with no TP53 mutation testing as TP53 disruptions.

Immunophenotyping and IGHV mutation analyses

Immunophenotyping of CD38 and ZAP-70 expression and IGHV sequencing were performed as described [28]. Positive cut-off values were 30% and 20%. Germline IGHV was defined as ≥98% germline homology.

Statistical analyses

Survival was calculated as time from diagnosis until death or last follow-up. Time-to-treatment (TTT) was calculated as time from diagnosis until first treatment. Calculations were performed using SPSS (version 19.0) software (IBM Corporation, Armonk, NY, USA). Categorical variables were compared using χ2 test and continuous variables using Student t-test. Survival curves were constructed by Kaplan-Meier method and log-rank test was used for significant associations. Multivariate analysis was done by Cox proportional hazard regression. Multivariate analysis of TTT was done. However, there were too few events to analyze survival. P-values were two-sided; P<0.05 was considered significant.