Targeted sequencing reveals the relationship between mutations and patients’ clinical indicators, blood cell counts and early progression in diffuse large-B cell lymphoma

Abstract

In the current study, we assessed the relationship between mutations and the blood cell counts and early progression of patients with diffuse large-B cell lymphoma (DLBCL). A total of 109 patients with newly diagnosed DLBCL were included in this study. UBE2A mutation was only found in patients with bone marrow involvement. The mutations of ZNF608, SF3B1, DTX1, and NCOR2 were related to blood cell counts. NCOR2 mutations were only detected in patients of the noncomplete response group (PR + SD + PD). In addition, the mutations of ATM, BTG2, TBL1XR1, and TP53 were linked to lower PFS/OS rate, while SGK1, SCOS1, and NFKBIE were related to higher PFS/OS rate. Importantly, we identified that Ann Arbor stage (III–IV), B symptoms, absolute lymphocyte count (ALC) abnormity, and MTOR mutation were the four independent influencing factors of the 12-month progression of DLBCL patients. Overall, this study revealed that mutations were associated with the early progression of DLBCL.

Keywords:

• Mutation
• blood cell counts
• response to therapy
• prognosis
• early progression
• diffuse large-B cell lymphoma

Introduction

Diffuse large-B cell lymphoma (DLBCL) is the most common type of non-Hodgkin lymphoma (NHL), accounting for 25–50% of NHL, with a higher proportion in developing countries than that of the developed countries. In China, about 25,000–30,000 new cases diagnosed with DLBCL annually, with the median onset age of 70 years [1]. DLBCL has been divided into two subgroups based on the gene expression profiling, the germinal center B-cell-like (GCB) subtype and the activated B-cell-like (ABC) subtype, remaining 10–15% unclassifiable [2]. Compared with the GCB subtype, patients with ABC subtype have an inferior outcome, for whom the 3-year progression-free survival (PFS) rate is approximately 40–50%, while it is 75% for the GCB subtype [3,4]. Although more than 60% DLBCL patients can be cured by R-CHOP immunochemotherapy (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone), 40% patients experience recurrent or refractory lymphoma for whom the prognosis was obviously worsened [5]. It has been reported that 5–10% DLBCL patients at early stage (I–II) and 20–25% patients at advanced stage (even achieved complete response (CR)) will develop to relapsed or refractory diseases [2]. Thus, it is important to find effective markers to accurately predict the early progression, aiming to improve the long-term prognosis.

Accumulated evidence has demonstrated that DLBCL patients always carry genetic alterations, as detected by the next-generation sequencing (NGS) [6,7]. The mutation spectra identified, however, differ somewhat in the published patient cohorts, indicting a high genetic heterogeneity in patients with DLBCL [8,9]. IGLL5, PIM1, CD70, TP53, and BTG2 are the most common mutated genes in Chinese DLBCL cohorts [10,11]. Noticeably, it has been demonstrated that some mutations are related to patients’ clinical features, prognosis, and treatment response. For example, Li et al. [12] found that NOTCH1 mutations were significantly associated with lower CR rate, PFS, and overall survival (OS) rates. MYD88 mutation was frequently detected in patients with multiple extranodal involvement, co-occurred with CD79B, PIM1, TBL1XR1, BTG1, MPEG1, and PRDM1 gene mutations, and correlated with multiple organ invasion, such as the bones, testes, kidney, breasts, skin, and ovaries [13]. CD79B and PIM1 mutations were associated with complex genetic events, double expression lymphoma (DEL), ABC subtype, advanced clinical stage, and poor prognosis in DLBCL [14]. Also, Ma et al. [14] revealed that the frequency of CD79B (42.86% vs. 9.38%, p = 0.000) and PIM1 mutations (38.78% vs. 17.71%, p = 0.005) showed a significant increase in patients with early progression (less than 12 months, POD12), and combination of CD79B mutations (OR = 5.970, p = 0.001), PIM1 mutations (OR = 3.021, p = 0.026) and the elevated lactate dehydrogenase (LDH) level (OR = 2.990, p = 0.018) was effective to predict the POD12 in R-CHOP treated DLBCL patients. However, the relationship between mutations and other clinical indicators, such as blood cell counts and the POD12 of patients responded to R-CHOP regimen still remains largely unknown.

In the present study, we assessed the mutation profile of patients with newly diagnosed DLBCL, and analyzed the relationship between mutations and patients’ clinical features, blood cell counts, response to therapy, prognosis, and POD12 in DLBCL patients received R-CHOP regimen.

Materials and methods

Patient information

Paraffin-embedded samples from 116 Chinese DLBCL patients were obtained from the Fourth Hospital of Hebei Medical University between September 2013 and September 2021. The concentration of seven samples was too low to sequence, and a total of 109 DLBCL patients with success sequencing data were included in this study. All of the 109 patients were newly diagnosed DLBCL and aged no less than 18 years old. All patients received R-CHOP-based standard treatment regimen following diagnosis and were followed up to January 2022. General clinical characteristics, including age, sex, Hans classification, Eastern Cooperative Oncology Group (ECOG), LDH level, International Prognostic Index (IPI), bulky disease, B symptom, β₂-microglobulin (B2M) level, number of extranodal involvement, bone marrow/peripheral blood involvement, Ann Arbor stage, response to therapy (CR; PR, partially response; SD, stable disease; PD, progressive disease), blood cell counts including absolute lymphocyte count (ALC), absolute monocyte count (AMC) and absolute neutrophil count (ANC), platelets (PLT), PFS, and OS were collected (Supplemental Table 1, Table 1). The study was approved by the ethic committee of the Fourth Hospital of Hebei Medical University.

Table 1. The clinical features of 109 patients with newly diagnosed DLBCL.

Clinical features	n (%)
Age (years)
≤60	57 (52.3)
>60	52 (47.7)
Gender
Female	60 (55.0)
Male	49 (45.0)
Hans classification
GCB	44 (40.4)
Non-GCB	64 (58.7)
Unknown	1 (0.09)
Extranodal involvement
0–1	83 (76.1)
≥2	24 (22.0)
Unknown	2 (1.8)
Bone marrow/peripheral blood involvement
Yes	18 (16.5)
No	88 (80.7)
Unknown	3 (2.8)
Ann Arbor
I–II	33 (30.3)
III–IV	76 (69.7)
B symptoms
Yes	44 (40.4)
No	65 (59.6)
Bulky disease
Yes	17 (18.9)
No	90 (82.6)
Unknown	2 (1.8)
IPI Score
0–2	59 (54.1)
3–5	46 (42.2)
Unknown	4 (3.7)
ECOG performance status
0–1	68 (62.4)
≥2	36 (33.0)
Unknown	5 (4.6)
LDH
Normal	56 (51.4)
Elevated	50 (45.9)
Unknown	3 (2.8)
B2M
Normal	60 (55.0)
Elevated	41 (37.6)
Unknown	8 (7.3)
ALC
Normal	67 (61.5)
Low	36 (33.0)
High	5 (4.6)
Unknown	1 (0.09)
AMC
Normal	85 (78.0)
Low	4 (3.7)
High	19 (17.4)
Unknown	1 (0.09)
ANC
Normal	87 (79.8)
Low	7 (6.4)
High	14 (12.8)
Unknown	1 (0.09)
PLT
Normal	87 (79.8)
Low	4 (3.7)
High	17 (15.6)
Unknown	1 (0.09)
Response to therapy
CR	53
PR	32
SD	4
PD	10
Unknown	10

ALC: absolute lymphocyte count; AMC: absolute monocyte count; ANC: absolute neutrophil count; B2M: β₂-microglobulin; COO: cell of origin; ECOG: Eastern Cooperative Oncology Group; IPI: International Prognostic Index; LDH: lactate dehydrogenase; PLT: platelets.

Targeted capture sequencing and data processing

Genomic DNA was extracted from the paraffin-embedded samples using a QIAamp Blood DNA Mini Kit (QIAGEN, Hilden, Germany) and 200 ng of genomic DNA was further used to prepare a captured library using Enzyme Plus Library Prep kit and TargetSeq One Kit (iGeneTech, Shanghai, China). Targeted sequencing was performed at iGeneTech Bioscience Company (Shanghai, China). The 114 gene panel is shown in Supplemental Table 2. Reads were aligned to the human genome reference assembly (UCSC Genome Browser hg19) and variants were annotated by ANNOVAR. Variants were filtered out if: (1) reads base quality <20; (2) variants on the positive-strand and negative-strand were inconsistent; (3) variant allele frequency (VAF) <10% and individual mutant reads < 20; (4) dbSNP (v147) sites that were not existed in COSMIC database; (5) variants frequency in 1KG project, ExAC and gnomAD databases >1%; (6) missense mutations were predicted not deleterious by Bioinformatics Tools (sift > 0.05, Polyphen2_HVAR_pred < 0.447, and CADD_phred <15); (7) mutations were predicted to be synonymous.

Statistical analysis

The Maftools (‘clinicalEnrichment’) R package was applied to assess the relationship between the mutation frequencies of genes and the clinical characteristics of DLBCL patients using Fisher’s exact tests, including gender, age (>60), Hans classification (GCB, non-GCB), extranodal involvement number (≥2), bone marrow/peripheral blood involvement, Ann Arbor stage (III–IV), B symptoms, bulky disease, IPI Score (3–5), ECOG (≥2), elevated LDH and B2M levels, deregulated AMC, ALC, PLT, and ANC levels. Genes with p values <0.05 were adjusted by Bonferroni’s correction. Kaplan–Meier’s (K–M) curves with log rank test were applied to analyze the relationship between.

To screen the independent influencing factors of the POD12 of patients with newly diagnosed DLBCL, clinicopathological characteristics (p< 0.05) and mutation genes (p< 0.1) with significant differences between 12-month progression and 12-month non-progression groups were included in the binary logistic regression analysis with a backward LR selection approach.

Results

Clinical features of DLBCL patients

Among the 109 patients, 44 cases (40.4%) were the GCB subtype and 64 cases (58.7%) were the non-GCB subtype according to the Hans classification. Fifty-seven (52.3%) patients aged ≤60 years and 60 (55.0%) patients were males. The proportion between different ages (≤60, >60), IPI Scores (0–1, and 2–5), B symptoms (yes, no), serum LDH levels (elevated, normal), and B2M levels (elevated, normal) were close. However, the proportion of patients with less extranodal involvement number (0–1; 76.1%), no bone marrow/peripheral blood involvement (80.7%), good ECOG performance status (0–1; 62.4%), non-bulky disease (82.6%), higher clinical stage (III–IV; 69.7%), and normal ALC (61.5%), AMC level (78.0%), ANC (79.8%), and PLT (79.8%) levels (Table 1) were reported.

Relationship between mutations and patients’ clinicopathological features

With the aid of the targeting sequencing of 114 lymphoma-related genes, we explored the mutation landscape of 109 patients with newly diagnosed DLBCL. A total of 1093 mutation sites in 100 genes were identified in this cohort, and ATM (28%), SAMHD1 (27%), MYD88 (26%), PIM1 (23%), IGLL5 (23%), SOCS1 (22%), KMT2D (22%), TP53 (18%), PCLO (17%), and JAK2 (17%) were the most common mutation genes, which were identified in at least 17 patients (Figure 1).

Figure 1. The mutation landscape of 109 patents with primary DLBCL. Top 50 mutated genes found in the 109 DLBCL patients.

Additionally, we assessed the relationship between the mutations and patients’ clinicopathological features. The results showed that SGK1 mutations were more common in the GCB subtype compared with the non-GCB subtype (20.5% vs. 3.1%) (Figure 2(A)). UBE2A mutations were only detected in patients with bone marrow involvement (Figure 2(B)). CHD2 mutations were more common in patients with B symptoms (15.9% vs. 1.5%) (Figure 2(C)). NCOR2 mutations were only detected in patients with low-medium IPI Score (0–2) (Figure 2(D)). The mutations of TMSB4X gene were more common in patients with elevated B2M (29.3% vs. 8.3%) (Figure 2(E)). These results demonstrated that the mutation profiles were significantly different between patients with different clinical features.

Figure 2. Assessment of the relationship between mutations and patients’ clinicopathologic features. The mutation frequencies of genes with significant differences between various items were determined, including (A) Hans. (B) Bone marrow involvement. (C) B symptom. (D) IPI. (E) β₂-microglobulin.

Mutations were associated with the blood cell counts of DLBCL patients

Hematologic markers, including the complete blood cells counts and its components are closely associated with the prognosis of malignancies [15]. For example, low level of ALC, a surrogate marker of inflammation and a marker of host immunity, have been demonstrated to be a poor prognostic factor in patients with DLBCL at diagnosis or after first relapse [16,17]. To clarify the relationship between mutations and blood cell counts, we assessed the relationship between the mutations and the ALC, AMC, ANC, and PLT levels in patients with DLBCL. The results showed that ZNF608 mutations were only discovered in patients with elevated ALC (Figure 3(A)). SF3B1 mutations were more common in patients with elevated ANC level (35.7% vs. 2.3%, 0.0%) (Figure 3(B)). In addition, the mutation frequencies of DTX1 were significantly increased in patients with low PLT level (50.0% vs. 10.3%, 0.0%) (Figure 3(C)). These results demonstrated that mutations were closely associated with DLBCL patients’ blood cell counts.

Figure 3. Assessment of the relationship between mutations and patients’ blood cell counts and response to therapy. The mutation frequencies of genes with significant differences between various items were determined, including (A) ALC. (B) ANC. (C) PLT. (D) Response to therapy.

NCOR2 mutations were associated with DLBCL patients’ response to R-CHOP based regimen

Also, we assessed whether the gene mutations were associated with patients’ response to R-CHOP based chemotherapy. Compared with patients achieved CR, the mutations of NCOR2 were only found in patients in the noncomplete response group (PR + SD + PD) (Figure 3(D)). To further explore the relationship between mutations and the response to R-CHOP regimen in DLBCL cohort, we also assessed whether the number of mutation sites/mutation genes and the mean VAF of mutations were different between patients achieved CR and PR + SD + PD followed R-CHOP based treatment. The results demonstrated that the number of mutation sites, the number of mutation genes and mean VAF showed no significant difference between patients with CR and PR + SD + PD (data not shown). These results revealed that NCOR2 mutations were closely associated with the response to therapy in DLBCL patients.

Mutations were associated with DLBCL patients’ prognosis

To clarify the correlation between mutations and prognosis in Chinese DLBCL patients, we analyzed the effects of gene mutation status and the PFS and OS of patients with DLBCL. The results showed that patients with ATM mutations have shorter PFS compared with that of patients without ATM mutations (Figure 4(A)), while DLBCL patients with SGK1 or SOCS1 mutations have longer PFS (Figure 4(B,C)). In addition, the OS of DLBCL patients with BTG2, TBL1XR1, or TP53 mutations was shorter than patients without these mutations (Figure 4(D,G,H)), while patients with NFKBIE or SGK1 mutation had longer OS (Figure 4(E,F)). These results revealed that the gene mutation status was closely associated with DLBCL patients’ prognosis.

Figure 4. Assessment of the relationship between gene mutations and patients’ prognosis. K–M curves were applied to analyze the relationship between the mutations of (A) ATM, (B) SGK1, and (C) SOCS1 and the PFS of patients with DLBCL, as well as the relationship between the mutations of (D) BTG2, (E) NFKBIE, (F) SGK1, (G) TBL1XR1, and (H) TP53 and the OS of patients with DLBCL.

MTOR mutations decreased the risk of POD12 in DLBCL patients

Also, we assessed the relationship between the clinical features and mutations with the POD12 of DLBCL patients. A total of 82 patients with a follow-up time of ≥12 months and five relapsed patients with a follow-up time of <12 months were included in this analysis, among which 26 patients were divided into the 12-month progression group and 61 patients were divided into the 12-month non-progression group. The proportion of patients achieved CR in the 12-month non-progression group was significantly higher than that of the 12-month progression group (p = 0.03), while higher Ann Arbor stage (III–IV) (p = 0.001), B symptoms (p = 0.002), and ALC abnormity (p = 0.001) were more common in the 12-month progression group. Other clinical features and blood cell counts showed no significant difference (p> 0.05), as summarized in Table 2.

Table 2. The clinical features between patients with 12-month progression and 12-month non-progression.

Clinical features	12-month progression (n = 26)	12-month non-progression (n = 61)	p
Age (years), n (%)			0.642
≤60	15 (57.7)	31 (50.8)
>60	11 (42.3)	30 (49.2)
Gender, n (%)			0.244
Female	11 (42.3)	35 (57.4)
Male	15 (57.7)	26 (42.6)
Hans classification, n (%)			0.470
GCB	8 (30.8)	25 (41.7)
Non-GCB	18 (69.2)	35 (58.3)
Unknown	0	1
Extranodal involvement, n (%)			0.141
0–1	18 (69.2)	50 (84.8)
≥2	8 (30.8)	9 (15.2)
Unknown	0	2
Bone marrow/peripheral blood involvement, n (%)			0.345
Yes	7 (26.9)	6 (15.4)
No	19 (73.1)	53 (84.6)
Unknown	0	2
Ann Arbor, n (%)			0.001
I–II	2 (7.7)	27 (44.3)
III–IV	24 (92.3)	34 (55.7)
B symptoms, n (%)			0.002
Yes	18 (69.2)	20 (32.8)
No	8 (30.8)	41 (67.2)
Bulky disease, n (%)			0.730
Yes	4 (15.4)	7 (11.9)
No	22 (84.6)	52 (88.1)
Unknown	0	2
ECOG, n (%)			0.200
0–1	13 (54.2)	42 (71.2)
≥2	11 (45.8)	17 (28.8)
Unknown	2	2
ALC, n (%)			0.001
Normal	10 (38.5)	46 (76.7)
Abnormal	16 (61.5)	14 (23.3)
Unknown	0	1
AMC, n (%)			0.158
Normal	18 (69.2)	50 (83.3)
Abnormal	8 (30.8)	10 (16.7)
Unknown	0	1
Absolute neutrophil, n (%)			1.000
Normal	21 (80.8)	49 (81.7)
Deregulation	5 (19.2)	11 (18.3)
Unknown	0	1
PLT, n (%)			0.341
Normal	20 (76.9)	52 (88.7)
Deregulation	6 (23.4)	8 (13.3)
Unknown	0	1
Number of mutation sites, n (%)			0.373
>Average (10)	3 (11.5)	13 (21.3)
≤Average (10)	23 (88.5)	48 (78.7)
Response to therapy, n (%)			0.030
CR	8 (34.8)	35 (61.4)
PR + SD + PD	15 (65.2)	22 (38.6)
Unknown	3	4

ALC: absolute lymphocyte count; AMC: absolute monocyte count; ANC: absolute neutrophil count; COO: cell of origin; ECOG: Eastern Cooperative Oncology Group; IPI: International Prognostic Index; PLT: platelets.

In addition, we compared the mutation features of the two groups. The proportion of mutation site number > average (10) in the 12-month progression group showed no significant difference to that of the non-progression group (p = 0.373). However, the mutation profiles of the two groups were different. PIM1 (35%), IGLL5 (31%), TP53 (27%), ATM (23%), MYD88 (23%), BTG2 (19%), KMT2D (19%), and MEF2B (19%) were the most frequent mutation genes in patients with 12-month progression, while SOCS1 (31%), SAMHD1 (31%), MYD88 (25%), PIM1 (25%), MTOR (23%), JAK2 (21%), CD79B (21%), and KMT2D (21%) were the most common mutations in patients from the 12-month non-progression group (Figure 5(A)). Moreover, the mutation frequency of MTOR and TNFAIP3 was significantly higher in patients of the 12-month non-progression group, while HLA-B mutations were only detected in patients of the 12-month progression group (Figure 5(B)).

Figure 5. Relationship between mutations and the 12-month progression in patients with DLBCL. (A) Top 50 mutated genes found in 87 DLBCL patients with disease progression information. (B) The mutation frequencies of genes with obvious differences between 12-month progression and 12-month non-progression groups were determined.

Importantly, we constructed a model to predict the POD12 of DLBCL patients using the binary logistic regression analysis. Clinical features, including Ann Arbor stage, B symptoms, ALC abnormity, and response to therapy which showed significant differences (p< 0.05), and the mutation status of HLA-B, MTOR, and TNFAIP3 genes with differences (p< 0.1) between the two groups were included. The results showed that Ann Arbor stage III–IV (OR, 6.934, 95% CI, 1.223–39.311, p = 0.029), B symptoms (OR, 5.943, 95% CI, 1.471–24.015, p = 0.012), ALC abnormity (OR, 3.693, 95% CI, 1.182–11.538, p = 0.025), and MTOR mutation (OR, 0.09, 95% CI, 0.008–0.954, p = 0.046) were the four independent risk factors of the POD12 of DLBCL patients (Table 3). These results highlighted the significant value of mutations in predicting the early progression in DLBCL patients.

Table 3. Multivariate Cox proportional hazard regression models for 12-month progression in patients with DLBCL.

Variable	Relative risk	95% CI	p
Ann Arbor stage (III–IV)	6.934	1.223–39.311	0.029
B symptoms	5.963	1.471–24.015	0.012
ALC	3.693	1.182–11.538	0.025
MTOR	0.09	0.008–0.954	0.046

ALC: absolute lymphocyte count.

Discussion

The widespread application of NGS has identified >100 cancer driver genes associated with lymphoma. Reassuringly, several genetic prognostic models have been established to predict the prognosis of DLBCL, such as the four and seven genetic classifications [10,18]. In the present study, we explored the association between mutations and patients’ clinical features, blood cell counts, response to therapy, prognosis, and early progression.

We found some gene mutations (VAF ≥10%) were specifically occurred in patents with specific characteristics. For example, UBE2A mutation was only found in patients with bone marrow involvement. UBE2A encodes a ubiquitination-related protein, and its mutation has been reported in about 10% Chinese DLBCL cases [8]. Here, we found that three patients with bone marrow involvement of the 109 DLBCL patients carried a UBE2A mutation with a VAF ≥10%, suggesting that mutation of UBE2A may provide an alternative method for the detection of bone marrow involvement in patients with DLBCL. In addition, we found that the mutation frequency of SGK1 was significantly higher in the GCB subtype as compared with the non-GCB subtype, which was consistent with a recent study [19]. Moreover, we found that CHD2 mutations were more commonly occurred in DLBCL patients with B symptoms. CHD2 encodes chromodomain helicase DNA binding protein 2 and appears to play a critical role in the development and hematopoiesis of tumors [20]. Chd2 mutated mice were reported to have increased extramedullary hematopoiesis and susceptibility to lymphomas, which suggested that the mutations in CHD2 gene may be linked to the occurrence of DLBCL [20]. Also, we observed that NCOR2 mutations were only detected in patients with low-medium IPI Score (0–2), which is associated with better therapy response and longer survival [21]. Thus, we speculated that NCOR2 mutations may also be linked to therapy response and survival, consistently, we found that NCOR2 mutations were not detected in patients achieved CR.

Accumulated evidence has demonstrated that the deregulation of blood cell counts is associated with the prognosis of DLBCL patients. The low peripheral blood ALC count [16,17], elevated AMC count [22], and high lymphocyte to monocyte ratio (LMR) [23], considered as surrogate markers for host immunity, have been found to be linked to the clinical outcomes of DLBCL patients, and thereafter improve the IPI-risk definition. Also, the low PLT count has been reported as a prognostic factor in hematological malignancies [24,25], including DLBCL [26]. Here, we first revealed the relationship between mutations and the numbers of ALC, ANC, AMC, and PLT. Our results showed that ZNF608 mutations were only detected in patients with high ALC, and SF3B1 mutations were more common in patients with high ANC, which may provide new targets for the treatment of DLBCL patients with deregulated ANC.

Gene mutations have been reported to be associated with DLBCL patients’ therapy response and prognosis. For example, NOTCH1 or TP53 mutations are linked to shorter OS [27], while SGK1 mutations are associated with longer OS in patients with DLBCL [19]. Consistently with many previous reports [28–33], we found that DLBCL patients with ATM, TP53, or BTG2 mutations have inferior outcome. In addition, we found that patients carrying NFKBIE mutation had a favorable prognosis. It has been demonstrated that NFKBIE mutations is associated with benign clinical outcomes in Burkitt’s lymphoma [34], while NFKBIE deletions are associated with poor outcome in primary mediastinal B‐cell lymphoma [35] and NFKBIE aberrations predict inferior outcome in chronic lymphocytic leukemia [36]. These contradictory conclusions may be explained by the different variant forms and the high heterogeneity of different diseases. SGK1 encodes a serine/threonine protein kinase that plays an important role in cellular stress response, and SGK1 mutated DLBCL patients are inclined to have better prognosis [19,27,30]. Consistently, our results showed that SGK1 mutated DLBCL patients had longer PFS and OS.

Accurate prediction may help us to early identify patients with early progression, leaving plenty of time to make a treatment plan. Here, we assessed the role of mutations combined with clinical features and blood cell counts in predicting the POD12 of DLBCL patients. Due to the limitation of included cases, seven factors were included in the binary logistic regression analysis, including Ann Arbor stage, B symptoms, ALC, therapy response and the mutations of HLA-B, MTOR, and TNFAIPS genes. Our results showed that, in addition to the Ann Arbor stage III–IV (p = 0.029), B symptoms (p = 0.012) and ALC abnormity (p = 0.025), MTOR mutation (p = 0.046) was the additional independent risk factor of the POD12 of DLBCL patients received R-CHOP regimen. MTOR encodes the conserved serine/threonine kinase mechanistic target of rapamycin kinase, a downstream effector of the PI3K/AKT pathway [37]. MTOR mutations, as well as other mutations in mTOR pathway genes, have been reported in DLBCL patients [38,39]. We first report that MTOR mutation is an independent risk factor of POD12 in DLBCL patients. Disease progression is less likely to occur in MTOR mutated DLBCL patients following the treatment of R-CHOP-based regimen, suggesting that DLBCL patients with MTOR mutations may more likely benefit from R-CHOP based therapy.

Several limitations need to be acknowledged to this study. The most important limitation lies in the fact that this was a retrospectively designed study without validation. Other limitation is that we did not include normal controls to filter the germline mutations. To reduce the impact, sift, Polyphen, COSMIC, and CADD databases were used to screen pathogenic variations.

Conclusions

This study revealed that gene mutations were significantly associated with the clinical features, blood cell counts, response to therapy, prognosis, as well as POD12 in DLBCL patients. Especially, UBE2A mutation was only detected in patients with bone marrow involvement, which may provide an alternative method for the detection of bone marrow involvement. MTOR mutation, as well as Ann Arbor stage III–IV, B symptoms, and ALC abnormity were four independent risk factors of the POD12 of DLBCL patients. This study enriches the knowledge about DLBCL.

Ethical approval

This study was performed in compliance with the Declaration of Helsinki, and approved by the Ethic Committee of the Fourth Hospital of Hebei Medical University.

Consent form

Informed consent was obtained from all individual participants included in the study.

Author contributions

GMZ and GYM designed the study. GYM, YHG, XTJ, CYH, HSL, XLW, ZG, YL, and SNZ provided the study material or patients in this study. GYM and YHG collected the data and performed the statistical data analysis. GMZ drafted and revised the manuscript. All authors contributed to the development of the manuscript and approved the final version.

Acknowledgements

We thank Shanghai Rightongene Biotechnology Co. Ltd (Shanghai, China) for the bioinformatics analysis.

Disclosure statement

The authors declare no conflict of interest.

Data availability statement

The data used to support the finding of this study may be released upon application to the corresponding author.

Additional information

Funding

This study was funded by the Beijing-Tianjin-Hebei Basic Research Cooperation Project (No. H2018206591).