This article is mentioned in:

Introduction

Diffuse large B-cell lymphoma (DLBCL) is a common histological form of non-Hodgkin's lymphoma (NHL), accounting for 30–40% of all adult NHLs (1,2). A diagnosis is made according to the morphology and immunophenotype of B cells (3). First-line management of DLBCL is a combination of chemotherapy drugs, rituximab, cyclophosphamide, adriamycin, vincristine and prednisone (R-CHOP) (4). The International Prognostic Index (IPI) (5) and age-adjusted International Prognostic Index (aaIPI) (6) serve important functions in daily practice to determine the treatment strategies and prognosis of individual cases. Neither the IPI nor the aaIPI identifies a risk group with <50% chance of survival (6). Thus, novel prognostic models are highly sought after. Inflammatory cells and cytokines located in tumors are more likely to contribute to cancer growth, spread, progression and immunosuppression than they are to mount an effective host antitumor activity (7–10). The inflammatory response can be represented by the level of serum neutrophils, lymphocytes, platelets, C-reactive protein and albumin (7) Previously, several combinations of these factors, including the neutrophil-lymphocyte ratio (NLR), lymphocyte-monocyte ratio (LMR) and platelet-lymphocyte ratio (PLR) have been reported to be useful prognostic factors in various malignant tumors (8,11–14). However, to date, no reports have investigated whether platelet count (PLT) and PLR are prognostic factors for DLBCL. The objective of this study was to investigate the prognostic ability of PLT and PLR in patients with DLBCL, and to obtain a novel prognostic scoring system for these metrics in order to predict the prognosis of the patients.

Patients and methods

Patients and clinicopathological variables

The clinical characteristics of 309 patients (including 186 males and 123 females) diagnosed with DLBCL between March 2009 and February 2015 at Tianjin Medical University Cancer Institute and Hospital (Tianjin, China) were retrospectively analyzed. All patients who were treated with R-CHOP-21 [rituximab (375 mg/m2) on day 1; cyclophosphamide (750 mg/m2), doxorubicin (50 mg/m2) and vincristine (1.4 mg/m2; maximal dose, 2 mg) on day 2; and prednisone (100 mg/day) on days 2 to 6] were diagnosed with DLBCL via pathological analysis. The dose and number of chemotherapy cycles were decided by physicians. The mean age of all patients was 58 years (range, 16–90 years), The median follow-up for the study was 47 months (range, 1–89 months). All patients included in the study were diagnosed with untreated DLBCL. Patients who had a previous history of malignancy, immunosuppression or previous treatment were excluded from the study. The available clinical parameters included age, gender, germinal center B-cell-like or non-germinal center B-cell-like disease, systemic B symptoms, Ann Arbor stage (15), IPI or aaIPI, hemoglobin (HGB), absolute lymphocyte count (ALC), absolute monocyte count (AMC), absolute neutrophil count (NEUT), PLT, serum albumin, serum lactate dehydrogenase level (LDH: Normal range 114–285 U/l), β2-microglobulin (β2m: Normal range 0.97–2.64 mg/l). The cell of origin was analyzed by immunohistochemistry. ALC, AMC, NEUT and PLT were obtained from pre-treatment CBC counts. According to IPI or aaIPI value, the patients were divided into four groups: i) Low-risk patients with IPI <2 scores or aaIPI=0; ii) low-intermediate risk patients with IPI=2 or aaIPI=1; iii) intermediate-high risk patients with IPI=3 or aaIPI=2; iv) high risk patients with IPI=4,5 or aaIPI=3. This study was approved by the ethics committee of Tianjin Medical University Cancer Institute and Hospital and was conducted in accordance with the Declaration of Helsinki.

Time from diagnosis of DLBCL to mortality from any cause was designated OS time. The time from diagnosis to lymphoma relapse, progression or mortality due to any cause was designated PFS time. The aim of the present study was to assess the effect of PLT, NLR, LMR and PLR on OS and PFS.

Statistical analysis

Statistical analysis was performed using SPSS software (version 17.0; SPSS, Inc., Chicago, IL, USA). Receiver operating characteristic (ROC) curves was used to determine the optimal threshold values of ALC, PLT, NLR, LMR and PLR. Patient characteristics were compared between PLR <170 vs. PLR ≥170 using the Pearson χ2 test. The survival curves were determined using the Kaplan-Meier method and the log-rank test. Multivariate analysis, which used Cox's proportional hazards model, was performed for the variables identified as statistically significant in univariate analysis to exclude confounding factors. A novel prognostic scoring system was then created to combine these factors, which were identified as statistically significant. P<0.05 was considered to indicate a statistically significant difference.

Results

Clinical characteristics of the patients with DLBCL

This study included 309 patients with newly diagnosed DLBCL. The mean age was 58 years (range, 16–90 years). A total of 186 patients were male, and 181/309 (58.6%) patients remained alive at the time of writing. Among the 128 mortalities, 123 individuals succumbed to disease recurrence, 4 individuals succumbed to infectious shock and 1 individual succumbed to respiratory failure. Baseline clinical and laboratory parameters are presented in Table I.

Table I. Patient characteristics.

Table I.

Patient characteristics.

Baseline characteristics	Patients, n (%)
Age
<60	178 (57.6)
≥60	131 (42.4)
Sex
Male	186 (60.2)
Female	123 (39.8)
B symptoms
Absent	209 (67.6)
Present	100 (32.4)
Ann Arbor stage
I–II	177 (57.3)
III–IV	132 (42.7)
IPI or aaIPI risk group
Low-risk	112 (36.2)
Low-intermediate risk	95 (30.8)
High-intermediate risk	56 (18.1)
High risk	46 (14.9)
Subtype
GCB	125 (40.5)
n-GCB	184 (59.5)
Albumin, g/l
≥35	259 (83.8)
<35	50 (16.2)
HGB, g/l
≥110	232 (75.1)
<110	77 (24.9)
ALC, ×109/l
≥1.45	172 (55.7)
<1.45	137 (44.3)
PLT, ×109/l
≥250	186 (60.2)
<250	123 (39.8)
NLR
≥2.9	125 (40.5)
<2.9	184 (59.5)
LMR
≥2.65	193 (62.5)
<2.65	116 (37.5)
PLR
≥170	165 (53.4)
<170	144 (46.6)
LDH (114–285 U/l)
>Normal	128 (41.4)
≤Normal	181 (58.6)
β2m (0.97–2.64 mg/l)
>Normal	131 (42.4)
<Normal	178 (57.6)

[i] IPI, international prognostic index; aaIPI, age-adjusted international prognostic index; GCB, germinal center B-cell like; nGCB, non-germinal center B-cell like; HGB, hemoglobin; ALC, the absolute lymphocyte count; PLT, platelet count; NLR, neutrophil-lymphocyte ratio; LMR, lymphocyte-monocyte ratio; PLR, platelet-lymphocyte ratio; LDH, lactate dehydrogenase; β2m, β2-microglobulin.

Selection of the best threshold values of ALC, PLT and PLR for DCBCL patients

ROC curve analyses determined the optimal threshold values for ALC, PLT and PLR. The optimal ALC threshold value was 1.45×109/l, with an area under the curve (AUC) value of 0.640 [95% confidence interval (CI), 0.578–0.703; P<0.001; Fig. 1A]. ROC curve analysis identified 250×109/l as the threshold value of PLT for predicting survival with an AUC of 0.622 (95% CI, 0.559–0.685; P<0.001; Fig. 1B). The threshold value of PLR was 170, with an AUC of 0.640 (95% CI, 0.577–0.703; P<0.001; Fig. 1C).

Figure 1. Receiver operating characteristic curve, AUC and 95% CI for the (A) absolute lymphocyte count, (B) platelet count and (C) platelet-lymphocyte ratio at diagnosis. AUC, area under curve; CI, confidence interval.

Association of PLR with the clinical characteristics in patients with DLBCL

ROC curve analyses determined the that the sum of the specificity and the sensitivity was the largest when PLR was 170, with an AUC of 0.640 (95% CI, 0.577–0.703; P<0.001). Therefore, a PLR of 170 was chosen as the threshold for division in to groups. Patients were divided into two groups: PLR <170 group or PLT ≥170 group. A total of 144 patients (46.6%) had a PLR <170 and 165 patients (53.4%) had a PLR ≥170. Compared with the patients with a PLR <170, the patients with PLR ≥170 were significantly associated with the presence of B syndromes, increased Ann-Arbor stages (15), high-intermediate risk or high risk (IPI ≥3, aaIPI ≥2), decreased albumin levels, decreased HGB levels and increased LDH. However, there was no significant association between PLR and age, gender, subtype or β2m (Table II).

Table II. Clinical comparison between patients with PLR<170 and PLR≥170.

Table II.

Clinical comparison between patients with PLR<170 and PLR≥170.

Baseline characteristics	PLR<170a, n	PLR≥170b, n	P-value
Age, years			>0.999
<60	83	95
≥60	61	70
Sex			0.728
Male	85	101
Female	59	64
B symptoms			<0.001
Absent	115	94
Present	29	71
Ann Arbor stage			<0.001
I–II	101	76
III–IV	43	89
IPI or aaIPI risk group			<0.001
Low-risk	75	37
Low-intermediate risk	43	52
High-intermediate risk	14	43
High-risk	12	33
Subtype			>0.999
GCB	59	67
n-GCB	85	98
Albumin, g/l			<0.001
≥35	137	122
<35	7	43
HGB, g/l			<0.001
≥110	132	100
<110	12	65
LDH (114–285 U/l)			<0.001
>Normal	39	89
≤Normal	105	76
β2m (0.97–2.64 mg/l)			0.728
>Normal	42	89
<Normal	102	76

a n=144

b n=165. IPI, international prognostic index; aaIPI, age-adjusted international prognostic index; GCB, germinal center B-cell like; n-GCB, non-germinal center B-cell like; HGB, hemoglobin; LDH, lactate dehydrogenase; β2m, β2-microglobulin.

Prognostic significance of PLT and PLR

Univariate analysis revealed that an age ≥60 years (P<0.001), the presence of B symptoms (P=0.001), stage III–IV disease (P<0.001), high-intermediate risk or high risk (P<0.001), decreased albumin levels (P<0.001), decreased HGB levels (P<0.001), ALC (P<0.001), NLR (P=0.019), LMR (P<0.001), increased LDH levels (P<0.001) and β2m (P<0.001) were poor prognostic factors (Table III). Patients with a PLT ≥250×109/l experienced a significantly decreased OS rate (5-year OS rate, 50.0 vs. 70.7%; P<0.001 Fig. 2A) and PFS (5-year PFS rate, 45.7 vs. 61.5%; P=0.003 Fig. 2B) than those with PLT<250×109/l. The OS and PFS rate in patients with a PLR ≥170 were significantly decreased compared with those with a PLR <170 at diagnosis (5-year OS rate, 41.8 vs. 77.1%, P<0.001 Fig. 2C; 5-year PFS rate, 35.8 vs. 70.6%, P<0.001, Fig. 2D). Furthermore, via multivariate analysis, age, IPI or aaIPI rick groups, β2m and PLR were identified to be independent prognostic factors for OS, whereas age, Ann-Arbor stage and PLR may independently predict poor PFS (Table IV).

Figure 2. Kaplan-Meier curve of (A) overall survival according to the PLT (P<0.001) and (B) progression free survival according to the PLT (P=0.003). Kaplan-Meier curve of (C) overall survival according to the PLR (P<0.001) and (D) progression-free survival according to the PLR (P<0.001). PLT, platelet count; PLR, platelet-lymphocyte ratio.

Table III. Analysis of prognostic factors for OS and PFS (univariate analysis) in 309 patients with DLBCL.

Table III.

Analysis of prognostic factors for OS and PFS (univariate analysis) in 309 patients with DLBCL.

Baseline characteristics	OS	P-value	PFS	P-value
Age, years		<0.001		<0.001
<60	68.5		61.8
≥60	44.3		38.5
Sex		0.836		0.715
Male	58.6		51.6
Female	57.7		52.5
B symptoms		0.002		0.001
Absent	63.2		57.2
Present	48.4		41.0
Ann Arbor stage		<0.001		<0.001
I–II	72.3		65.5
III–IV	39.4		33.6
IPI or aaIPI risk groups		<0.001		<0.001
Low-risk	77.7		72.3
Low-intermediate risk	62.1		51.6
High-intermediate risk	36.8		29.8
High risk	28.9		29.5
Subtype		0.552		0.155
GCB	62.1		58.9
n-GCB	55.5		47.0
Albumin, g/l		<0.001		<0.001
35	62.9		57.0
<35	34.0		26.1
HGB, g/l		<0.001		<0.001
≥110	63.8		57.6
<110	41.6		35.1
ALC (×109/l)		<0.001		<0.001
≥1.45	69.8		64.3
<1.45	43.8		36.5
PLT (×109/l)		<0.001		0.003
≥250	50.0		45.7
<250	70.7		61.5
NLR		0.019		0.003
≥2.9	51.2		43.2
<2.9	63.0		57.9
LMR		<0.001		<0.001
≥2.65	66.3		60.4
<2.65	44.8		37.9
PLR		<0.001		<0.001
≥170	41.8		35.8
<170	77.1		70.6
LDH (114–285 U/l)		<0.001		<0.001
>Normal	43.2		63.5
≤Normal	69.1		35.4
β2m (0.97–2.64 mg/l)		<0.001		<0.001
>Normal	39.7		33.6
≤Normal	71.9		65.5

[i] IPI, international prognostic index; aaIPI, age-adjusted international prognostic index; GCB, germinal center B-cell like; n-GCB, non-germinal center B-cell like; HGB, hemoglobin; ALC, the absolute lymphocyte count; PLT, platelet count; NLR, neutrophil-lymphocyte ratio; LMR, lymphocyte-monocyte ratio; PLR, platelet-lymphocyte ratio; LDH, lactate dehydrogenase; β2m, β2-microglobulin.

Table IV. Multivariate analysis of prognostic factors for survival in testing set.

Table IV.

Multivariate analysis of prognostic factors for survival in testing set.

	Overall survival		Progression-free survival
Characteristics	P-value	HR (95% CI)	P-value	HR (95% CI)
Age	<0.001	2.410 (1.678–3.460)	<0.001	2.022 (1.488–2.823)
B symptoms	0.348	1.209 (0.814–1.797)	0.185	1.277 (0.889–1.834)
Ann Arbor stage	0.123	1.441(0.906–2.294)	0.038	1.575 (1.026–2.418)
IPI or aaIPI risk groups	0.009	1.402 (1.088–1.806)	0.161	1.183 (0.935–1.495)
Albumin	0.413	1.231 (0.748–2.026)	0.478	1.183 (0.743–1.885)
HGB	0.354	0.807 (0.513–1.269)	0.627	0.900 (0.589–1.375)
PLT	0.221	1.321 (0.846–2.064)	0.852	0.961 (0.636–1.454)
ALC	0.080	1.472 (0.955–2.270)	0.067	1.456 (0.937–2.178)
NLR	0.314	0.815 (0.548–1.213)	0.764	0.945 (0.652–1.369)
LMR	0.139	0.758 (0.525–1.095)	0.083	0.738 (0.523–1.040)
PLR	<0.001	0.327 (0.205–0.521)	<0.001	0.418 (0.274–0.640)
LDH	0.240	1.294 (0.842–1.989)	0.036	1.536 (1.028–2.295)
β2m	0.043	1.499 (1.012–2.219)	0.133	1.327 (0.917–1.920)

[i] OS, overall survival; PFS, progression-free survival; IPI, international prognostic index; aaIPI, age-adjusted international prognostic index; GCB, germinal center B-cell like; n-GCB, non-germinal center B-cell like; HGB, hemoglobin; ALC, the absolute lymphocyte count; PLT, platelet count; NLR, neutrophil-lymphocyte ratio; LMR, lymphocyte-monocyte ratio; PLR, platelet-lymphocyte ratio; LDH, lactate dehydrogenase; β2m, β2-microglobulin; CI, confidence interval.

Prognostic significance of the PLR and β2m combined with IPI or aaIPI

As the diagnosis and treatment of DLBCL has improved, the ability of IPI or aaIPI to differentiate between risk groups, particularly high-risk groups, has declined (Fig. 3A-D). Thus, according to the results of the Cox regression analysis, a novel score was created combining the PLR and β2m with IPI or aaIPI, 309 patients were split into three groups: i) Low-risk patients with a PLR <170, IPI <2 scores or aaIPI=0 and normal β2m; ii) high-risk patients with a PLR ≥170, IPI ≥4 or aaIPI=3 and high level of β2m; and iii) intermediate-risk patients. This novel score predicted a 5-year OS rate of 86.4, 54.1 and 21.1% in the low-, intermediate- and high-risk groups, respectively (P<0.001; Fig. 3E). The estimated 5-year PFS rate with this stratification was: 81.4% for the low-risk group, 47.0% for the intermediate-risk group and 21.1% for the high-risk group (P<0.001; Fig. 3F).

Figure 3. Kaplan-Meier curve of (A) OS times according to IPI risk group (P<0.001) and (B) PFS times according to IPI risk group (P<0.001). Kaplan-Meier curve of (C) OS times according to aaIPI risk group (P<0.001) and (D) PFS times according to aaIPI risk group (P<0.001). Kaplan-Meier curve of (E) OS times according to risk group combined with PLR and β2M (P<0.001) and (F) PFS times according to risk group combined with PLR and β2M (P<0.001). OS, overall survival; PFS, progression free survival; IPI, international prognostic index; aaIPI, age-adjusted international prognostic index; PLR, platelet-lymphocyte ratio; β2M, β2-microglobulin.

Discussion

DLBCL is a highly aggressive NHL with varied clinical manifestations and variable patient prognosis. IPI is the widely accepted prognostic factor index for patients with DLBCL (5). The addition of rituximab to conventional chemotherapy has led to a marked improvement in patient survival rates (4). As a result, the ability of IPI or aaIPI to differentiate between risk groups, particularly high-risk groups, has declined (5). The revised International Prognostic Index is unable to identify a risk group with a <50% chance of survival (6). This means that more sensitive prognostic factors are required. Inflammation is recognized as a major hallmark of cancer. As early as 1863, Rudolf Virchow, who suggested that lymphoreticular infiltration reflected the origin of cancer at sites of chronic inflammation, identified a connection between inflammation and cancer (16). Subsequently, numerous studies have provided evidence that the host inflammatory responses serve a critical function in various aspects of cancer, including cancer initiation, promotion, progression and metastasis (7,17–19) Previously, a number of inflammatory markers have been proposed to have potential for use as predictors of OS and PFS for solid tumors (8,12,20–23). The present study identified that NLR, LMR and PLR may be able to predict the prognosis of patients with DLBCL.

Monocytes and neutrophils, which are important components in the active defense system, are potent regulators of macrophages, mast cells and epithelial cells, and serve an important function in inflammatory events (16). These cell types are able to differentiate into tumor-associated macrophages in tumor tissue, which undergo tumor-promotion and M2-like macrophage polarization and secrete angiogenic factors, including interleukin-8, vascular endothelial growth factor (VGEF) and fibroblast growth factor, then induce further tumor angiogenesis and progression (24,25). Furthermore, monocyte-derived cells may provide nutritional factors that directly promote the growth and survival of malignant tumors (17–19).

Lymphocytes are the basic components of the immune system; they can induce cytotoxic cell death and produce cytokines in cancer cells (26,27). Lymphocytopenia impairs the antitumor immune response of the host, which in turn promotes tumor expansion and leads to a poor patient prognosis.

Tumors require the formation of new blood vessels to provide nutrients and oxygen for continued lesion growth. Platelets release VEGF upon their activation, thus promoting angiogenesis (28). Platelet activation also protects tumor cells from natural killer cells, and platelets support spontaneous metastasis (22–23). Platelet-derived lysophosphatidic acid enhances bone metastatic growth and progression, and platelet-released factors may serve a function in tumor progression (29). Therefore, these results provide evidence that platelet counts may serve as inflammatory biomarkers and could aid the prediction of the prognosis of patients.

β2m is a human leukocyte antigen-class I molecule that is expressed on the membrane of almost all nucleated cells. Studies have reported that β2m, which regulates p21-activated kinases, VEGF and fatty acid synthase-mediated growth and survival signaling pathways, is a growth factor and signaling molecule in cancer cells (30,31); it serves a critical function in the proliferation, apoptosis and metastasis of cancer cells (30,32). Furthermore, β2m is a useful prognostic factor in multiple types of cancer, including breast (33), gallbladder (34), prostate and lung cancer (31), chronic lymphocytic leukemia (35) and follicular lymphoma (36). In the present study, an increased level of β2m was associated with poor patient prognosis in DLBCL.

In summary, PLT and PLR were associated with poor prognosis in patients with DLBCL. The NLR and LMR at diagnosis, as biomarkers combining an estimate of host immune and tumor microenvironment, were previously hypothesized to be powerful prognostic factors in patients with newly diagnosed DLBCL (37); the results of the present study confirmed this. Furthermore, it was identified that PLR was an independent predictor of survival in patients who were newly diagnosed with DLBCL. Additionally, it was identified that the patients who have a poor prognosis may be divided through a novel prognostic scoring system (PLR and β2m combined with IPI or aaIPI), which is of great significance for the evaluation of prognosis and guiding treatment.

Acknowledgements

The authors would like to thank Dr Xiaofang Wang for encouragement and guidance. The present study was funded by the National Natural Science Foundation of China (grant no. 81272562).

Glossary

Abbreviations

Abbreviations:

References