Diagnostic Value of 18F-PET/CT in Patients with Primary Lymphoma and the Prognostic Value of Maximum Standardized Uptake Value (SUVmax) in Diffuse Large B-Cell Lymphoma (DLBCL)

The diagnostic accuracy of 18 F-PET/CT was assessed in patients with primary lymphoma and the clinical application value of SUV max was determined. The diagnostic accuracy of a total of 97 patients with initial 18 F-PET/CT scans between January 2015 and February 2020 were assessed, and the SUV max was compared according to the different pathological subtypes. The relationship between SUV max and immunophenotype, clinical characteristics, and genetic types were estimated. According to the pathological results, 10 cases were misdiagnosed by PET/CT, and the accuracy was about 90%. Statistical analysis did not reveal a signicant difference between Hodgkin’s lymphoma (HL) and non-Hodgkin’s lymphoma (NHL) (p = 0.9071). Among NHL, the average SUV max was statistically different between diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (FL) (p = 0.0004), FL and natural killer/T-cell lymphoma (p = 0.0078), FL and peripheral T-cell lymphoma (PTCL) (p = 0.0117), DLBCL and mantle cell lymphoma (MCL) (p = 0.0294). In patients with DLBCL, SUV max was correlated with the expression level of proliferation index Ki-67 (r = 0.33, p = 0.018), while average SUV max shows no difference between various immunophenotype expression levels, ages, gender, skeletal invasion situations, clinical grade stages, international prognostic index (IPI) score, and different gene types (germinal center B cell-like (GCB) and non-GCB).


Background
A lymphoma is a group of malignant tumors originating from lymph nodes or extranodal lymphoid tissues with high heterogeneity and classi cation complexity. 18-uorodeoxyglucose positron emission tomography/computed tomography ( 18 F-FDG PET/CT) has high clinical value in staging and re-staging, evaluation of curative effect, follow-up after treatment, and prognosis in lymphoma [1,2]. 18 F-FDG PET/CT can effectively identify areas that are missed or lymphoma lesions that are misclassi ed by CT alone. Although many researches have studied the application of 18 F-FDG in lymphoma, most of them focus on the application in grade or prognosis evaluation in one speci c lymphoma subtype [3][4][5].
However, the use of 18 F-FDG PET/CT imaging for the analysis of the metabolic activity (uptake of FDG) difference among various lymphoma subtypes have not yet been addressed adequately. The uptake of FDG may be different in lymphomas of different pathological subtypes, as well as the same pathological subtypes with different immunophenotypes, Although some previous studies considered that the level of 18 F-FDG uptake in lymphoma lesions is related to the grade and expression level of some immunophenotypic molecules of the tumor, the phenomenon is yet controversial and deserves further research [5][6][7][8]. For example, ki67 was con rmed to be related to SUV max in some studies, but in some studies it was proved to be unrelated [4,9,10]. In order to reveal the performances for the variable 18 F-FDG uptake in different types of lymphoma lesions and the value of uptake in judging the prognosis of lymphoma, we compared and analyzed the 18 F-FDG uptake in lymphoma lesions among 97 patients with newly diagnosed lymphoma and evaluated the accuracy of PET-CT diagnosis. Consecutively, the difference in uptake among different pathological subtypes was assessed. In addition, we explored the correlation between uptake value and clinical features, prognosis index [international prognostic index (IPI) factors], gene expression and different immunophenotypes of patients with diffuse large B-cell lymphoma (DLBCL).

Patients
We collected data on lymphoma patients, who underwent 18 F-FDG PET-CT imaging at the Shanghai Chest Hospital, Shanghai Changzheng Hospital and Dongfang Hospital from January 2015 to February 2020.
According to the criteria for inclusion and exclusion, 97 patients with NHL were enrolled in this study, including 50 males and 47 females, aged 18-85 (54.3 ± 16.9) years. A total of 5 cases were Hodgkin's lymphoma (HL), and the remaining were non-Hodgkin's lymphoma (NHL). Among non-Hodgkin's lymphoma, 57 cases were diffused large B-cell lymphoma (DLBCL), 8 were follicular lymphomas (FL), 7 were natural killer (NK)/T-cell lymphomas, 8 were peripheral T-cell lymphomas (PTCL), 4 were mucosaassociated lymphoid tissue (MALT), 4 were mantle cell lymphomas (MCL), and the remaining 4 were diagnosed as NHLs but the speci c pathological type not clari ed. According to the result of pathology, all the patients were enrolled according to the following criteria: (1) diagnosed by pathology, and detailed immunohistochemical results were included in the diagnosis; (2) untreated; (3) PET-CT imaging and pathological examination did not exceed 4 weeks as assessed by the same method; (4) the highest FDG uptake area in the lesion was coherent to that in the biopsy or operation area. Patients, who had undergone surgery or received chemotherapy, were excluded. Table 1 summarizes the characteristics of all included patients. 2. 18 F-FDG PET/CT imaging and interpretation 2.1 Imaging method All patients underwent staging 18 F-FDG PET/CT before any treatment (local surgery or chemotherapy). In both hospitals, Philips GXL 16 PET-CT (Philips Medical Systems, Inc., Cleveland, OH, USA) was used as the imaging device. The imaging agent 18 F-FDG was provided by Shanghai Atomic Kexing Pharmaceutical Co., Ltd. with radiochemical purity >95%. 18 F-FDG was injected after at least 6 h fasting and the glucose level <10 mmol/L. The dose of 18 F-FDG was 3.70-5.18 MBq/kg, and the images were acquired at 60±10 min after the injection. Written consent was obtained from all hospitals before the study.

Detection of metabolism activity of lymphoma
The SUV max of lymphoma was obtained by the average of 4-8 consecutive layers of lesions on PET images. For every layer, a manual region of interest (ROI) over the area of maximum activity was drawn, and SUV max was estimated as the highest SUV of the pixels within the ROI. In the case of single lesions, the SUV max was measured directly, while for multiple lesions, the highest SUV max value in the whole body was considered as the SUV max value of the patient, and if the intake was negative value, the SUV max value of the largest lesion in the whole body was considered as the SUV max value of the patient.

Immunophenotype, clinical information, and gene information
Immunohistochemistry was used for detecting the samples. The specimens were xed in 10% neutral buffer formaldehyde solution, embedded in para n, and sliced into 4-μm-thick sections. The staining was performed for molecules Bcl-6, Bcl-2, CD10, CD23, Mum-1, Pax-5, Ki67, CD2, CD3, CD5, EMA, CD138, CD30, and ALK, and the results were determined based on the number of positively stained cells recorded. The results were recorded as follows: negative (-): <10%; weakly positive (+): 10-30%; moderately positive (++): 30-75%; strongly positive (+++): >75%; an expression of ≥10% is considered positive. The clinical characteristics of the patients, including age (according to the age division regulation of WHO, the patients were divided into youth group with age <44 years, middle-age group with age 45-59 years, and elderly group with age >60 years), gender (male and female), tumor clinical grade stage (stage I/II was divided into low-grade group and stage III/IV was classi ed into the high-grade group), presence of bone metastasis, and IPI score (low-risk, IPI 0-2 or aaipi 0-1 and high-risk, IPI 3-5 or aaipi≥2) were assimilated. Consecutively, the genotype of the patients was assessed (GCB or non-GCB).

Statistical analysis
All statistical analyses were conducted using SAS 9.2 software (SAS Institute, Carey, NC). The statistical signi cance of SUV max between pathological subtype groups, different levels of molecule expression groups and different gene expression groups was analyzed by Fisher's exact test or Student's t-test and Wilcoxon two-sample test. The correlation between SUV max and Ki-67 was evaluated by Pearson's correlation test. p<0.05 was considered statistically signi cant.

Accuracy of PET-CT diagnosis
A total of 10 diagnostic errors were detected among all the enrolled patients, of which, 2 were misdiagnosed with the negative uptake of PET-CT; the pathological subtypes were MALT and DLBCL ( Fig. 1), and SUV max values were 0.8 and 2.1, respectively. The remaining 8 cases were misdiagnosed as high uptake of PET-CT. The pathological subtypes included 5 cases with DLBCLN, 1 with FL, 1 with classic HL, and 1 with NHL, whose speci c subtype could not be determined. All cases of missed diagnosis occurred in the gastrointestinal tract and the detail information are showed in Table 2. 3.2 Analysis of the metabolic activity of different lymphoma subtypes Different lymphoma subtypes have signi cantly different metabolic activities. Among all included patients, the highest lesion was measured in the abdominal lymph nodes of a patient with DLBCL with SUV max value of 51. (Fig. 2). Based on the pathological type of grouping, the average SUV max value was calculated for each pathological type. The average SUV max of HL was 16.8 ± 5.3, and that of NHL was 17.0 ± 5.1; among these, the value for DLBCL was 18.9 ± 4.9, FL was 7.8 ± 1.4; PTCL was 16.4 ± 4.9; NK/Tcell lymphoma was 18.7 ± 2.6; MALT was 13.7 ± 4.6, and MCL was 9.2 ± 3.4. The detail information was list in Table 3. Statistical analysis did not reveal any signi cant difference between HL and NHL (p = 0.9071). Among NHL, the mean SUV max differed signi cantly between DLBCL and FL (p = 0.0004), FL and NK/T-cell lymphoma (p = 0.0078), FL and PTCL (p = 0.0117), and DLBCL and MCL (p = 0.0294) (Fig. 3A).
3.3 Relationship between major clinical indications, immunophenotype molecular expression, and gene expression in DLBCL and SUV max As DLBCL has the highest incidence in China, the largest proportion of cases would allow further exploration of the correlation between SUV max and the expression of various immune markers in DLBCL.

Discussion
PET-CT is critical for the detection of lymphoma, several previous studies have proved that the application of PET-CT plays a major role in the diagnosis, assessment of treatment outcome, and prognosis of lymphoma [11][12][13]. Juweid et al. carried out the Imaging Subcommittee of International Harmonization Project, which suggested PET-CT as a routine monitoring method in the treatment of lymphoma in the future. Although the utility of PET/CT in lymphoma has been continually studied, few studies have focused on its misdiagnosis rate and missed diagnosis, as well as the analysis of the characteristics of these cases. Therefore, among nearly 100 cases of lymphoma, we were concerned about the rate of misdiagnosis of PET-CT. In the current study, a nal con rmation by pathology retrieved 10/97 lymphoma cases that were either misdiagnosed or missed in the initial PET-CT scan. Our multicenter study showed that the misdiagnosis rate of PET/CT was approximately 10.3%; 2/10 patients were missed diagnosis and negative on PET-CT with no uptake. Both cases were lymphomas of gastrointestinal system that were proved to be MALT by pathology, which agrees the ndings of  [16][17][18][19]. This phenomenon prompted the present study and based on our results, it seems that 18 F-FDG PET/CT was not recommended in these patients who are suspected of having gastric MALT lymphoma. In future studies, we will collect additional cases to verify the PET-CT ndings of the gastrointestinal MALT. The other 8 cases were misdiagnosed, half of these were localized in the gastrointestinal tract, putatively due to the speci c FDG uptake pattern of lymphoma in the gastrointestinal tract and this phenomenon is consistent with many previous studies that primary gastrointestinal lymphoma is a speci c type of digestive system tumor and its FDG uptake pattern is more likely to have low intake due to physiologic FDG activity in the gastrointestinal tract although variability in the degree of uptake occurred in various histologic subtypes of primary gastrointestinal lymphoma [17,20,21]. Hwang et al. also con rmed that SUV max can be used as a prognostic marker for gastrointestinal lymphoma as it differed markedly from other gastrointestinal cancers and is one of the reliable differential diagnostic criteria [22,23]. However, the disease is insidious in onset and still easy to be misdiagnosed and missed in the examination; In this study, two of the misdiagnosed cases were misdiagnosed as in ammation, while the remaining were misdiagnosed as other tumors. Thus, although the application of PET-CT has greatly improved the diagnosis rate of lymphoma, it is still necessary to focus on the characteristic manifestations of lymphoma and differentiate it from other benign lesions or other tumors.
The de nite diagnosis of different lymphoma subtypes plays a key role in the subsequent treatment and prognosis of the patient and several previous studies have focused on SUV max , but the value of PET-CT in the differential diagnosis of lymphoma subtypes remains controversial [24][25][26]. In this study, the main subtypes of NHL and HL were assessed, and the SUV max of each pathological subtype was estimated.
The current study showed that although no signi cant difference was observed in HL and NHL, the SUV max value of FL was signi cantly lower than other subtypes, and varied signi cantly from DLBCL, PTCL, and NK/T cell lymphoma. Moreover, a signi cant difference was detected between DLBCL and MCL (p = 0.0294), indicating that PET-CT is not only useful in detecting and nding the lesions but also provides information in the differential diagnosis of lymphoma. Thus, help determine the pathological subtypes of lymphoma and develop an appropriate treatment plan, which is considered signi cant.
The incidence of DLBCL in China is > 40% with respect to NHL, which is similar to our collected cases. We aspire to further analyze the correlation between the clinical indicators and prognostic indicators with SUV max in DLBCL as the immunophenotype is critical for determining follow-up treatment plan for different patients; The results showed that SUV max was positively correlated with the expression of Ki-67 in B cells. Although there have been several studies on Ki-67 in lymphoma, whether SUV max is related to Ki-67 is yet controversial. For example, studies of Storto et al. and Novelli et al. showed that the Ki-67 expression did not related to SUV max in T-cell NHL and MALT [5,8], while the current study was similar to that of Watanabe et al. [27], which proved that SUV max on 18 F-FDG-PET was associated with Ki-67 expression level and re ected the tumor aggressiveness in NHL. Ki-67 is a proliferative cell-associated nuclear antigen [28], and its function is closely related to mitosis. It is a popular biological indicator used in the investigation of many malignant tumors which can re ect the proliferation of malignant tumor cells. Therefore, our data showed that SUV max may can judge the malignant degree and prognosis of Bcell NHL.
In the current study, no signi cant difference was found in the SUV max values with respect to other grouping indicators, including age, sex, bone metastasis, IPI score, clinical grade, genotype, and other immunohistochemical indexes. Of these, the signi cant correlation between SUV max and IPI is yet controversial worldwide. Our ndings were consistent with those of Ding et al. and Adams et al. [29,30] that no signi cant correlation was established between SUV max and IPI. However, some other studies found that SUV max was associated with revised-IPI and the progression-free survival (PFS), and predicted the survival outcome in lymphoma patients [31,32]. The possible reasons for the discrepancy between the results of this study and some previous studies is that the subtypes of lymphoma are different. We only focused on DLBCL and the expression of immune molecules of different subtypes is quite different, as well as the SUV max value. Furthermore, the number of temporarily included cases may not be enough to re ect these relationships. Thus, we will increase the number of cases in future studies to further verify the signi cance of SUV max in DLBCL.

Conclusion
Although 18 F-PET/CT has signi cant diagnostic value in patients with primary lymphoma, some misdiagnosis occurs; hence, it is recommended to combine the clinical features and pathology when diagnose especially when diagnosing primary gastric lymphoma. Besides, SUV max is valuable in the differential diagnosis of different pathological types of NHL. More importantly, the SUV max of patients with DLBCL correlated with Ki-67 re ect that it may be a promising prognostic and e cacy indicator for invasiveness in patients with DLBCL.

Consent for publication
All patients signed written informed consent for the use of their data in this Article.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.