This article is mentioned in:

Introduction

Non-Hodgkin's lymphoma (NHL) is a group of malignant lymphohematopoietic tumors mainly occurring in lymphatic organs, including lymph nodes, spleen and thymus. In Asia, NHL frequently displays high grade and aggressive subtypes, which indicates a relatively poor prognosis (1). Currently, histopathology is the most important basis for the diagnosis of NHL (2,3); however, clinicians are usually required to formulate therapeutic strategies before pathological results are available. Therefore, it is necessary to identify patients with poor prognosis and provide them with more comprehensive therapies. Certain blood biochemical indices, including serum inflammatory cytokines, tumor burden indicators, and proportion of absolute lymphocyte to absolute monocyte (4), are of interest to clinicians. However, some clinicians still prefer radiological or pathological examinations since they reflect a more intuitive progress of the disease (5). Whether these non-serological examinations are superior to blood biochemical indices is controversial (6,7). Therefore, it is necessary to perform meticulous comparisons of these indicators.

As the most precise radiographic method to evaluate tumor metabolic activity, 18F-fludeoxyglucoase (FDG) positron emission tomography/computed tomography (PET/CT) is widely applied in tumor staging, grading and therapeutic assessment. The maximal standardized uptake value (SUVmax) is a descriptive parameter to indicate the extent of 18F-FDG uptake in normal and tumor tissue, and usually positively correlated with tumor malignancy (8–10). The predictive value of SUVmax in solid tumor has been demonstrated in several studies (11–13), whereas it remains elusive whether SUVmax is instructional in hematologic malignancies. Therefore, further studies are required to determine the role of SUVmax in hematologic malignancies.

Ki67, also known as antigen Ki67 or MKi67, is associated with cellular proliferation and is only expressed in proliferating cells. The ratio of Ki67 (Ki67%) in tumor cells prior to and after chemotherapy is of significance in medication guidance, particularly for the selection of cell-cycle-specific chemotherapeutic drugs (14). At present, no definitive conclusion has been made as to whether Ki67% has sufficient accuracy in predicting the prognosis of patients with NHL (15,16). Therefore, further studies are required.

The present study provided an elaborate assessment of the relationship among serological factors, SUVmax and Ki67%, as well as their predictive value regarding the long-term prognosis of patients with NHL.

Materials and methods

Patients and outcomes

Cases of newly-diagnosed NHL at Shanghai Tongji Hospital (Shanghai, China) between July 2015 and March 2019 were collected and evaluated for age, sex, pathological diagnosis, staging, grading, invasiveness and post-chemotherapy effectiveness. These patients also underwent necessary serological, radiological and pathological examinations. The inclusion criteria were defined as follows: i) The pathological diagnosis was clear and uncontroversial; ii) the patient had received regular chemotherapy and followed up for at least 2 months and iii) at least one radiological or pathological examination was performed. Some patients lack staging or grading information during hospitalization, but with complete serological, radiological and pathological data. For those patients, the data were also considered analyzable. Following removal of information-deficient individuals, 120 patients with valid clinical data were included in the present study. Diagnosis of these patients was performed according to the World Health Organization classification for tumors of hematopoietic and lymphoid tissue (17). The International Prognostic Index (IPI) was calculated before treatment to estimate the prognosis of patients with NHL (18). The patients were staged according to Ann Arbor Staging Classification (19). The patients received regular chemotherapy [cyclophosphamide, doxorubicin, vincristine and prednisone (CHOP); or rituximab-CHOP] and were assessed for effectiveness after four courses of treatment. The criteria of efficacy were defined as follows: i) Complete remission (CR), all clinical and radiological lesions had disappeared, lymph nodes and lumps shrunk to normal size, or the sum of products of greatest diameters (SPD) was reduced by >75%, tumor markers and biochemical indicators were in the normal range and stable for ≥4 weeks; ii) partial remission (PR), SPD is reduced by >50%, no new lesion appeared and maintained for ≥4 weeks; iii) stable disease, non-PR and non-progressive disease (PD); and iv) PD, the size of any abnormal lymph node determined prior to treatment increased by >50% compared with the previous SPD minimum, or a new lesion appeared during or after treatment. Survival data were also collected in order to determine prognosis-associated factors. Patients were followed up every 3 weeks during chemotherapy and every 3 months after discharge. The last follow-up was in September 2019. Overall survival (OS) time was defined as the interval from trial admission to death of any cause, and progression-free survival (PFS) time was defined as the interval from trial admission to disease progression or death from NHL.

Serological examinations

Serological factors included inflammatory cytokines and tumor burden indicators. The inflammatory cytokines were C-reactive protein (CRP), tumor necrosis factor α (TNF-α), interleukin-6 (IL-6), interleukin-2 receptor (IL-2R) and interleukin-8 (IL-8), whereas the tumor burden indicators were lactate dehydrogenase (LDH), ferritin and blood β2-microglobulin (β2-mg). TNF-α, IL-6, IL-2R and IL-8 were detected using chemiluminescence with Siemens interleukin test kits (IMMULITE and IMMULITE 1000 TNF-α/IL-6/IL-2R/IL-8; cat. no. LKNF1, LK6P1, LKIP1 and LK8P, respectively.) provided by Siemens AG. Fasting venous blood was obtained from patients on the first day of hospitalization and automatically tested with an Immulete 1000 Analyzer (Siemens AG). β2-mg and ferritin were detected using a radioimmunoassay. CRP was detected using immunoturbidimetry and LDH was measured using a creatine kinase assay. The normal ranges for each indicator were: CRP, <10 mg/l; TNF-α, <8.1 µg/l; IL-6, <5.9 ng/l; IL-2R, 223–710 U/ml; IL-8, <62 ng/l; LDH, 120–250 U/l; ferritin, 30–400 µg/l (male), 13–150 µg/l (female); β2-mg, 0.7–1.8 mg/l.

Radiological examination

All patients signed informed consent forms before PET/CT examination. The patients were required to fast for ≥6 h prior to intravenous infusion of 18F-FDG. Fasting glucose was measured and a dose of 18F-FDG ranging between 4.1 and 10.7 mCi was injected into patients depending on their weight and glucose level. After injection, patients rested for ≥40 min in a comfortable position and PET/CT scanning was performed from the skull to the proximal femur. The images were generated through Siemens Biograph Turepoint 40 PET/CT Imager and reconstructed after attenuation correction. The images were merged through Xeleris software. Subsequently, the reconstructed images were assessed by two experienced nuclear medicine physicians and SUVmax was calculated after the range of interest was drawn. Lesions were diagnosed according to the criteria recommended by the International Group for Imaging Diagnosis of Lymphoma (20). In order to describe the association between SUVmax and Ki67%, two different types of SUVmax were calculated, including SUVmax of the whole body (SUVmax-overall) and SUVmax of the biopsy site (SUVmax-biopsy). Only incipient PET/CT data were collected, and data after chemotherapy or surgery were automatically excluded.

Pathological examination

Immunohistochemical staining was performed in order to determine Ki67%. For pretreatment steps, the specimens were fixed with 10% neutral formaldehyde at room temperature for 24–48 h, embedded in paraffin wax, cut into 3-µm thick sections and stained with hematoxylin at room temperature for 5 min and eosin at room temperature for 1 min. Hematoxylin and eosin were provided by Solarbio Science & Technology (Beijing) Co., Ltd. For Ki67 immunohistochemical staining, the paraffin sections were stained using Ki67 antibody (cat. no. GT2094; 1:100; Gene Tech Biotechnology Co., Ltd.) on an automated immunostainer (cat. no. GAS95; Gene Tech Biotechnology Co., Ltd.). The antigen was retrieved by heating with EDTA (pH 9.0) at 100°C for 20 min. Blocking was performed using 3% H2O2 at 37°C for 10 min. The primary antigen was incubated at 37°C for 30 min. EnVision reagent (HRP anti-Rabbit/Mouse, cat. no. GK801030, Gene Tech Biotechnology Co., Ltd.) was incubated at 37°C for 20 min. The sections were stained by diaminobenzidine (DAB) at room temperature for 8 min, and then stained with hematoxylin at room temperature for 5 min. The nuclei stained with antibody were considered positive, and the number of Ki67-positive tumor cells in 1,000 tumor cells was calculated by 2 experienced pathologists under a light microscope at the 200× magnification. Examples of stained Ki67 sections were shown in Fig. S7. PBS was used as a negative control and Ki67-positive sections were regarded as the positive control.

Statistical analysis

All clinical data were analyzed using SPSS v20.0 software (IBM Corp.). Continuous variables are presented as mean ± SD, categorical variables are presented as n (%). The association between two different indicators was assessed by Spearman's non-parametric correlation analysis. For survival analysis, Kaplan-Meier curves were drawn and the log-rank test was used to compare the survival between groups. χ2 and P-values were calculated to show the extent of significance for survival data. Univariate and multivariate Cox regression analysis was performed to determine prognosis-associated indicators. The regression model was obtained with the method of ‘Forward: Wald’ (21). Receiver operating characteristic (ROC) curves were depicted to identify the optimal cut-off for indicators, and the area under the curve (AUC) as well as 95% CI were calculated to determine their predictive value. Youden index (sensitivity + specificity-1) was calculated to find optimal cutoff for those indicators. Mann-Whitney test was used to compare the levels of OS- or PFS-predicting indicators between patients that did or did not achieve CR. P<0.05 was considered to indicate a statistically significant difference.

Results

Patient characteristics

Among 120 patients aged between 15 and 93 years, 77 (64.2%) were male and 43 (35.8%) were female, and their respective average ages were 59±16 and 62±16 years. The age of all patients, whether grouped by sex or not, was normally distributed. A total of 86 (71.7%) patients were diagnosed with B-cell NHL (B-NHL) and 34 (28.3%) patients were diagnosed with T-cell NHL (T-NHL). Among them, 104 (86.7%) were diagnosed with invasive NHL, and 16 (13.3%) were diagnosed with non-invasive NHL. According to Ann Arbor stage (n=112), most patients were in stage IV (46; 41.1%), and the numbers of patients in stage I, II and III were 5 (4.5%), 22 (19.6%) and 39 (34.8%), respectively. A total of 55 (49.1%) patients had B symptoms (fever, night sweat or weight loss), whereas the other 57 (50.9%) patients did not exhibit B symptoms (n=112). A total of 44 (41.5%) patients reached CR after four regular courses of chemotherapy, whereas 62 (58.5%) patients did not reach CR (n=106). The median follow-up time for all patients was 14.5 months (range, 0.6–43.6 months). Details of patient characteristics are provided in Table I.

Table I. Characteristics of 120 patients with NHL.

Table I.

Characteristics of 120 patients with NHL.

Characteristic	Value
Median age, years (range)	61 (15–93)
Sex, n (%)
Male	77 (64.2)
Female	43 (35.8)
Type, n (%)
B-NHL	86 (71.7)
DLBCL	62 (51.7)
FL	11 (9.2)
MCL	6 (5.0)
MALT	5 (4.2)
Burkitt	2 (1.7)
T-NHL	34 (28.3)
AITL	12 (10.0)
PTCL	10 (8.3)
NK/T	4 (3.3)
TLBL	4 (3.3)
ALCL	4 (3.3)
Invasiveness, n (%)
Invasive	104 (86.7)
Non-invasive	16 (13.3)
Ann Arbor stage, n (%)a
I	5 (4.5)
II	22 (19.6)
III	39 (34.8)
IV	46 (41.1)
B symptoms, n (%)a
Yes	55 (49.1)
No	57 (50.9)
CR after treatment, n (%)b
Yes	44 (41.5)
No	62 (58.5)

a n=112

b n=106. Data are presented as the median (range) or number (percentage). NHL, non-Hodgkin's lymphoma; DLBCL, diffuse large B-cell lymphoma; FL, follicular lymphoma; MCL, mantle cell lymphoma; MALT, mucosa-associated lymphoid tissue lymphoma; AITL, angioimmunoblastic T-cell lymphoma; PTCL, peripheral T-cell lymphoma; NK/T, natural killer/T-cell lymphoma; TLBL, T lymphoblastic lymphoma; ALCL, anaplastic large cell lymphoma; CR, complete remission.

Correlation analysis

In order to explore the association among different categories of indicators, correlation analysis was performed. This unveiled an association between inflammatory cytokines and tumor burden indicators. LDH was correlated with CRP, IL-6, IL-2R and TNF-α (P<0.001). β2-mg and ferritin also exhibited a significant correlation with these inflammatory cytokines (P<0.05). However, there was no significant correlation identified between IL-8 and any of the aforementioned tumor burden indicators (P>0.05). Additionally, LDH, β2-mg, IL-6 and IL-2R were associated with age (P<0.05; data not shown). Furthermore, LDH, β2-mg, CRP, IL-6, IL-2R and TNF-α exhibited a significant correlation with stage (P<0.05; data not shown). By comparison, there was no significant correlation identified between SUVmax-overall and any of the serological factors (P>0.05), the same conclusion for SUVmax-biopsy. A noticeable correlation was identified between Ki67% and ferritin (Ρ=0.235; P=0.020). The correlation between SUVmax and Ki67% was also significant, and it appeared that SUVmax-biopsy exhibited a stronger association with Ki67% (Ρ=0.529; P<0.001) than SUVmax-overall (Ρ=0.395; P=0.017; Fig. 1; Table II).

Figure 1. Correlation analysis for SUVmax and Ki67% in 120 patients with non-Hodgkin's lymphoma. (A) Correlation between SUVmax-overall and Ki67%; (B) Correlation between SUVmax-biopsy and Ki67%. Compared with SUVmax-overall, SUVmax-biopsy displayed a better correlation with Ki67% (Ρ=0.529 for SUVmax-biopsy, compared with Ρ=0.395 for SUVmax-overall). Ki67%, ratio of Ki67; SUVmax, maximal standardized uptake value.

Table II. Correlations among serological factors, SUVmax and Ki67% in 120 patients with NHL.

Table II.

Correlations among serological factors, SUVmax and Ki67% in 120 patients with NHL.

	CRP		IL-6		IL-2R		IL-8		TNF-α		SUV max-overall		SUV max-biopsy		Ki67%
Indicators	ρ	P-value	ρ	P-value	ρ	P-value	ρ	P-value	ρ	P-value	ρ	P-value	ρ	P-value	ρ	P-value
LDH	0.507	<0.001c	0.557	<0.001c	0.608	<0.001c	0.076	0.483	0.638	<0.001c	0.055	0.739	0.068	0.692	0.065	0.504
β2-mg	0.522	<0.001c	0.551	<0.001c	0.669	<0.001c	0.045	0.688	0.660	<0.001c	0.170	0.352	0.238	0.206	−0.203	0.052
Ferritin	0.506	<0.001c	0.452	<0.001c	0.338	0.002b	0.011	0.918	0.260	0.016a	0.262	0.141	0.319	0.080	0.235	0.020a
SUVmax-overall	0.114	0.507	0.165	0.359	−0.119	0.510	−0.157	0.382	0.072	0.692					0.395	0.017a
SUVmax-biopsy	0.068	0.703	−0.017	0.929	−0.121	0.518	−0.169	0.365	−0.081	0.666					0.529	<0.001c
Ki67%	−0.031	0.761	0.076	0.500	−0.074	−0.511	0.148	0.187	−0.183	0.103	0.395	0.017a	0.529	<0.001c

a P<0.05

b P<0.01

c P<0.001. CRP, C-reactive protein; IL-6, interleukin-6; IL-2R, interleukin-2 receptor; IL-8, interleukin-8; TNF-α, tumor necrosis factor-α; LDH, lactate dehydrogenase; β2-mg, β2-microglobulin; SUVmax, maximal standardized uptake value; Ki67%, ratio of Ki67; NHL, non-Hodgkin's lymphoma.

Since different subgroups may exhibit different levels of serological, radiological or pathological indexes, a subgroup analysis based on T-/B-NHL or invasive/non-invasive NHL was performed. Compared with when all patients were included, the association between LDH and TNF-α in patients with T-NHL became more significant, but most indicators demonstrated an attenuated correlation (Table SI). For patients with B-NHL, the correlations were similar to the results when all patients were included in the study. Notably, SUXmax-overall was no longer significantly correlated with Ki67%, whereas the correlation between SUXmax-biopsy and Ki67% retained its significance (Table SI). When only patients with invasive NHL were included in the analysis, most correlations remained similar (Table SII). Regarding individuals with non-invasive NHL, the sample size was too small for correlation analysis. Therefore, this was not performed.

Survival analysis

To estimate the predictive value of different indicators, patients were automatically stratified into two groups. For serological indicators (CRP, TNF-α, IL-6, IL-2R, IL-8, LDH, ferritin and β2-mg), the patients were divided into groups according to the normal range of each indicator, with a normal group and an elevated group; For SUVmax and Ki67%, patients were divided into groups according to the median value. Based on the Kaplan-Meier survival curves, it became apparent that CRP had the greatest capacity to predict OS of patients with NHL (χ2=10.124, P<0.001). LDH, IL-6 and IL-8 were also prognostic factors (χ2=9.325, P=0.002; χ2=4.968, P=0.026; and χ2 =8.507, P=0.004, respectively, Fig. 2), whereas β2-mg, ferritin, IL-2R, TNF-α, SUVmax and Ki67% were not statistically significant in OS prediction (data not shown). The subgroup analysis further demonstrated the significance of IL-8 in OS estimation of patients with T-NHL (P=0.038; Fig. S1). Furthermore, CRP, LDH and IL-6 had the potential to predict OS of patients diagnosed with B-NHL (P<0.01; Fig. S2). For aggressive NHL, CRP, LDH and IL-8 were observed to be significant indicators (P<0.05; Fig. S3), whereas no analysis was performed for non-aggressive NHL due to insufficient sample size.

Figure 2. Survival curves of (A) CRP, (B) LDH, (C) IL-6 and (D) IL-8 in overall survival estimation of 120 patients with non-Hodgkin's lymphoma (P<0.001, P=0.002, P=0.026 and P=0.004, respectively). CRP, C-reactive protein; IL, interleukin; LDH, lactate dehydrogenase.

For PFS, a similar result highlighting the predictive value of CRP (χ2 =11.463; P<0.001), LDH (χ2=7.689; P=0.006) and IL-6 (χ2=4.366; P=0.037) was observed (Fig. 3). The subgroup analysis revealed that SUVmax-biopsy was a significant prognostic factor for patients with T-NHL (χ2=5.266; P=0.022; Fig. S4). However, none of the other indicators were statistically significant in patients with T-NHL, probably due to limited sample size and heterogeneity of data. By contrast, CRP, LDH and IL-6 had the ability to predict PFS of patients diagnosed with B-NHL, and IL-6 appeared to distinguish PFS of patients with B-NHL more remarkably compared with when all patients were included (χ2=7.327; P=0.007; Fig. S5). For aggressive NHL, only CRP was useful for PFS prediction (χ2=9.632; P=0.002; Fig. S6), and patients with non-aggressive NHL were automatically excluded due to small sample size.

Figure 3. Survival curves of (A) CRP, (B) LDH and (C) IL-6 in progression-free survival estimation of 120 patients with non-Hodgkin's lymphoma (P<0.001, P=0.006 and P=0.037, respectively). CRP, C-reactive protein; IL, interleukin; LDH, lactate dehydrogenase.

Univariate and multivariate analysis

Since multiple variables were included in the present study, a univariate analysis followed by a multivariate analysis was performed to determine their predictive value. In Cox regression analysis, multiple variables, including age, sex, aggressiveness, type, stage, B symptoms, IPI score, SUVmax, Ki67% and all serological factors were set as independent variables. To perform univariate analysis, each indicator was added to the variable list individually. When entry was set at 0.05 and removal at 0.10, with the method of ‘Forward: Wald’, significant indicators, including age (P=0.003), type (P=0.019), stage (P=0.006), B symptoms (P=0.004), IPI score (P<0.001), LDH (P=0.001), β2-mg (P<0.001), ferritin (P=0.015), CRP (P<0.001) and IL-2R (P=0.019), were obtained for OS estimation (data not shown). Subsequently, multivariate analysis was performed with the same method and the aforementioned indicators. A regression model containing LDH (P=0.007), β2-mg (P=0.002), CRP (P=0.012) and IL-8 (P=0.001) was obtained (Table III), indicating that these indicators were independent prognostic factors.

Table III. Cox regression analysis for overall survival estimation of 120 patients with non-Hodgkin's lymphoma.

Table III.

Cox regression analysis for overall survival estimation of 120 patients with non-Hodgkin's lymphoma.

Variables	Β	SE	Wald	P-value	RR (95% CI)
LDH	0.001	0.000	7.169	0.007	1.001 (1.000–1.002)
β2-mg	0.403	0.131	9.421	0.002	1.496 (1.157–1.934)
CRP	0.024	0.010	6.358	0.012	1.024 (1.005–1.044)
IL-8	0.037	0.011	11.025	0.001	1.037 (1.015–1.060)

[i] The regression model contains only variables screened from the univariate analysis. LDH, lactate dehydrogenase; β2-mg, β2-microglobulin; CRP, C-reactive protein; IL-8, interleukin-8; SE, standard error; RR, relative risk.

For PFS estimation, univariate analysis revealed age (P=0.011), type (P=0.004), stage (P=0.004), B symptoms (P=0.001), IPI score (P<0.001), LDH (P=0.001), β2-mg (P<0.001), ferritin (P=0.015) and CRP (P<0.001) as significant indicators (data not shown), and the multivariate analysis demonstrated that LDH (P=0.001), β2-mg (P=0.003), CRP (P=0.003) and IL-8 (P=0.042) were significant prognostic indicators (Table IV).

Table IV. Cox regression analysis for progression-free survival estimation of 120 patients with non-Hodgkin's lymphoma.

Table IV.

Cox regression analysis for progression-free survival estimation of 120 patients with non-Hodgkin's lymphoma.

Variables	Β	SE	Wald	P-value	RR (95% CI)
LDH	0.001	0.000	11.683	0.001	1.001 (1.000–1.002)
β2-mg	0.211	0.070	9.110	0.003	1.235 (1.077–1.416)
CRP	0.018	0.006	8.796	0.003	1.018 (1.006–1.030)
IL-8	0.012	0.006	4.142	0.042	1.012 (1.000–1.023)

[i] The regression model contains only variables screened from the univariate analysis. LDH, lactate dehydrogenase; β2-mg, β2-microglobulin; CRP, C reactive protein; IL-8, interleukin-8; SE, standard error; RR, relative risk.

ROC curves

The predictive accuracy of these significant prognostic factors was further evaluated using ROC curves. This revealed that β2-mg had the highest potential to predict OS of patients with NHL (AUC, 0.785; 95% CI, 0.690–0.881; P<0.001; Fig. 4). CRP (AUC, 0.741; 95% CI, 0.639–0.843; P<0.001) and LDH (AUC, 0.728; 95% CI, 0.629–0.826; P<0.001) were also indicated to be prognostic factors for OS. As for IL-8, ROC curves did not indicate any significant differences (P=0.145; data not shown), and this was therefore not further discussed. By calculating the Youden Index (sensitivity + specificity-1), the optimal cut-off was determined for these prominent indicators. The most suitable cut-off for β2-mg with the optimal effectiveness of prediction was 2.89 mg/l (sensitivity, 0.960; specificity, 0.467), The best cut-off value for LDH to predict OS was 215 U/l (sensitivity, 0.900; specificity, 0.489). For CRP, the Youden Index reached the maximum when the cut-off was set at 12.01 mg/l (sensitivity, 0.821; specificity, 0.427) (data not shown).

Figure 4. Receiver operating characteristic curves of (A) β2-mg, (B) CRP and (C) LDH in overall survival estimation of 120 patients with non-Hodgkin's lymphoma. β2-mg, β2-microglobulin; CRP, C-reactive protein; LDH, lactate dehydrogenase.

As for PFS, a similar result demonstrating the predictive potential of β2-mg (AUC, 0.712; 95% CI, 0.608–0.815; P<0.001), CRP (AUC, 0.700; 95% CI, 0.599–0.800; P<0.001) and LDH (AUC, 0.689; 95% CI, 0.587–0.792; P<0.001) was observed (Fig. 5), and the optimal cutoff points were set as 2.78 mg/l (sensitivity, 0.882; specificity, 0.530), 12.01 mg/l (sensitivity, 0.757; specificity, 0.589) and 224 U/l (sensitivity, 0.775; specificity, 0.538), respectively.

Figure 5. Receiver operating characteristic curves of (A) β2-mg, (B) CRP and (C) LDH in progression-free survival estimation of 120 patients with non-Hodgkin's lymphoma. β2-mg, β2-microglobulin; CRP, C-reactive protein; LDH, lactate dehydrogenase.

CR prediction of OS or PFS indicators

Further investigations were performed to identify whether these serological indicators predicting OS or PFS could predict CR as well. Mann-Whitney U tests demonstrated that patients who reached CR after four rounds of chemotherapy exhibited significantly lower levels of CRP and β2-mg at the beginning of treatment compared with those patients who did not achieve CR (P<0.001; data not shown), indicating that CRP and β2-mg were prognostic factors for survival and CR evaluation. Among the other indicators in survival analysis, IL-6 was also a prognostic factor in CR assessment (P<0.001; data not shown), providing evidence that serological factors have a higher potential to predict CR after four rounds of chemotherapy compared with SUVmax and Ki67%.

Discussion

With the advantage of intuitively displaying size, location and metabolic activity of malignancies, PET/CT has been widely applied for precise staging and therapeutic evaluation. A higher SUV is indicative of a larger quantity of vigorous and invasive tumor tissue, and a corresponding poor prognosis (8–10). Several studies have demonstrated this relationship in NHL (22,23). In the present study, only a limited significance for SUVmax-biopsy to predict PFS of patients with T-NHL was observed. This may be due to heterogeneity of data, and may not represent the actual situation. A more detailed subgroup analysis based on individual diseases, such as DLBCL, is required for future studies in order to reduce bias. Furthermore, some articles (24,25) have suggested substitution of SUVmax with a more responsive indicator, the alteration of SUVmax (∆SUVmax), to evaluate interdependency. However, since repeated PET/CT examination may cause a large financial burden and additional exposure to radiation, not all patients consent to frequent follow-up examinations and clinical data may be difficult to collect. One solution is to calculate SUVmax of the biopsy site (SUVmax-biopsy) instead of the entire body (26). This has been demonstrated in the present study and certain other studies, and a more significant correlation between SUVmax-biopsy and Ki67% was observed.

Ki67% is considered to be one of the most useful pathological indicators to reflect tumor proliferation (27). However, in the present study, it was not a significant prognostic factor for NHL. One primary reason is that the definition of Ki67% is largely subjective and technology-dependent. At times, pathologists tend to overestimate the grade when the percentage of Ki67-positive tumor cells is high. Additionally, the location of the biopsy site affects judgement, and low-quality samples may increase measurement errors and reduce the efficacy of correlation assessment. Previous studies do not provide affirmative answers regarding whether Ki67% has promising predictive power as opposed to other mainstream predictors (6,28). Perhaps a combination of Ki67% with other pathological indicators will provide a more suitable option. However, it is certain that the association between SUVmax and Ki67% is close, which indicates that the metabolic activity of the tumor is associated with its proliferative activity, and this has been demonstrated in multiple studies (10,29).

The present study indicated that traditional serological factors exhibited a higher predictive value than SUVmax and Ki67%. In the survival analysis, CRP was demonstrated to be the best predictor for both OS and PFS. Additionally, LDH and IL-6 had a marked practicality in the estimation of OS and PFS. By performing Cox regression analysis, it was demonstrated that CRP, LDH and β2-mg were fully independent predictors of OS and PFS, and their ROC curves indicated high sensitivity and specificity. Furthermore, CRP, β2-mg and IL-6 were significant predictors of CR after four courses of chemotherapy. This is instructional because clinicians may regard them as key indicators for NHL prognosis and formulate individualized therapies based on their detection level. However, the present study also had some limitations. First, the sample size was relatively small. An increased sample size and extended observation of patient outcome are required in future studies. Second, the number of cases exhibited a great variation among different subgroups. As a result, certain subgroups were too small to be analyzed. Finally, PET data may not be sufficiently adequate to draw firm conclusions. However, our study have demonstrated that serological factors are superior to SUVmax and Ki67%, which are of clinical importance and have a guidance value.

In conclusion, the present study provided comparisons of serological, radiological and pathological indicators and their ability to predict survival and treatment outcomes for patients with NHL. Overall, it is recommended to use β2-mg, CRP and LDH for estimating OS and PFS as the first choice, and taking CRP, β2-mg and IL-6 into account for CR prediction. CRP and β2-mg can be used for estimating survival as well as CR. By contrast, SUVmax and Ki67% did not exhibit sufficient predictive power and should not be recommended. The present study may serve as a reference for clinicians to formulate therapeutic regimens for patients diagnosed with NHL by referring to indicators with high predictive value.

Supplementary Material

Supporting Data

Acknowledgements

The authors would like to thank Dr Bing Xiu (Department of Hematology, Shanghai Tongji Hospital, Shanghai Tongji University School of Medicine), for her guidance in writing this article, and her support for statistical analysis and polishing the language of the manuscript. The authors would also like to thank Dr Yu Zeng, Dr Bing Li, Miss Yihan Liu and Miss You Wu (Department of Hematology, Shanghai Tongji Hospital, Shanghai Tongji University School of Medicine), who joined in this project and provided reasonable suggestions and necessary support in writing the manuscript.

Funding

This article was supported by grants from the National Natural Science Foundation (grant nos. 81600156 and 81401882), the clinical research cultivation project of Shanghai Tongji Hospital (ITJ(QN)1907) and the Key Project of Natural Science Foundation of China (grant no. 81830004).

Availability of data and materials

The datasets generated and/or analyzed during the present study are available from the corresponding author on reasonable request.

Authors' contributions

JL collected PET/CT data and was a major contributor in writing the manuscript. YZ provided all necessary pathological data including supplementary immunohistochemical photographs. YW was the main collector of serological data and contributed to writing the manuscript. BX revised the manuscript and provided assistance for statistical analysis. BL and XL collected basic information of enrolled patients and collected survival data from patient follow-up. BX, JF, WZ and AL contributed to the conception and design of this study as well as acquisition of funding. All authors have read and approved the final manuscript.

Ethics approval and consent to participate

All experiments were ethically approved by The Ethics Committee of Shanghai Tongji Hospital (approval no. KYSB-2016-19), and written informed consent was provided by each patient.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References