Efficacy of 18F-2-fluoro-2-deoxy-D-glucose Positron Emission Tomography/Computerized Tomography for Bone Marrow Infiltration Assessment in the Initial Staging of Lymphoma

Objective: Currently 18F-2-fluoro-2-deoxy-D-glucose (18F-FDG) positron emission tomography/computerized tomography (PET/CT) is being successfully used for staging and follow-up of Hodgkin’s lymphoma (HL) and non-Hodgkin’s lymphoma (NHL). Various studies have demonstrated that PET/CT effectively detects bone marrow involvement (BMI) and is concordant with bone marrow biopsy (BMB) findings, thus it is deemed as a complementary method. This study was aimed to evaluate 18F-FDG-PET/CT efficiency for detection of BMI in HL and NHL. Methods: The study included 172 lymphoma cases who were admitted to Akdeniz University Medical School Department of Nuclear Medicine for initial staging with PET/CT. Visual and semiquantitative assessments were performed for PET/CT scan findings of the cases. The maximum standard uptake (SUVmax) value was the quantitative parameter used for 18F-FDG-PET scan. In visual assessment, bone marrow metabolic activity that is greater than the liver was considered as pathologic. For semiquantitative assessment, regions of interest were drawn for SUVmax estimation, which included iliac crest in cases with diffusely increased metabolic activity and the highest activity area in cases with focal involvement. BMB was considered as the reference test. Results: On visual assessment of all the cases, PET/CT was found to yield 31% sensitivity and 85% specificity rate for detection of BMI. On visual assessment of HL cases, sensitivity rate was determined as 80%, and specificity as 78%, while in NHL cases the corresponding values were 24% and 90%, respectively. On semiquantitative assessment of HL cases, considering SUVmax≥4, sensitivity was found as 80% and specificity as 68%. In NHL patients, considering SUVmax≥3.2, sensitivity rate was detected as 65% and specificity as 58%. Conclusion: In this study, a moderately high concordance was observed between PET/CT and BMB findings. PET/CT appears to be a significant method for detecting BMI. Although PET/CT is not a substitute for BMB, we suggest it can be used as a guide to biopsy site and a complementary imaging technique for BMB.


Introduction
Accurate staging of lymphomas is essential both to implement effective treatment protocols and minimize side effects (1). Identification of bone marrow infiltration (BMI) has an important role in staging (2). Bone marrow involvement indicates generalized disease in lymphoma patients, and the standard method established for its evaluation is bone marrow biopsy (BMB). BMB from unilateral iliac crest is the routine first-line method used for staging (3,4). However, this method has certain limitations since it is an invasive method and allows for evaluation of a limited part of the bone marrow. BMI can also be detected by imaging techniques. Computerized tomography (CT) detects cortical bone lesions and late stage bone changes. However, it has a low sensitivity rate for detecting early stage changes (5). Magnetic resonance (MR) is not used in routine practice since it is a sensitive but costly technique, which needs longer imaging time and is anatomically limited. Currently, 18 F-2-fluoro-2-deoxy-D-glucose ( 18 F-FDG) positron emission tomography (PET/CT) is being successfully used for both staging and follow-up of Hodgkin's lymphoma (HL) and non-Hodgkin's lymphoma (NHL) (2,6,7). Various studies have demonstrated that PET/CT effectively detects bone marrow involvement and is concordant with BMB findings (1,5,8). Thus, it is deemed as a complementary method (9). This study aimed to evaluate the efficacy of 18 F-FDG-PET/CT in detection of BM infiltration in HL and NHL.

Patients
This study, approved by the Akdeniz University Medical School Local Ethics Committee, included histopathologically confirmed, treatment naïve 172 lymphoma cases (50 F, 122 M; age interval 3-85; mean age 45.37±21.14; 64 HL, 108 NHL) who underwent initial staging with PET/CT at Akdeniz University Medical School Department of Nuclear Medicine between July 2009 and December 2013. Patients included in the study did not have other concomitant malignancies. Additionally, patients did not receive any bone marrow stimulation therapy before PET/CT scanning. The maximum interval between PET/CT scan and BMB was 10 days.

Positron Emission Tomography/Computerized Tomography Scanning
Intravenous 0.1 mCi/kg 18 F-FDG was administered to each patient following 6 hours of fasting, with a blood glucose level below 200 mg/dL. The intravenous/oral contrast agent was administered. After 45-60 minutes of waiting period, PET/CT images were acquired from the vertex to the upper thigh with Siemens Biograph True Point PET/CT scanner (CT section thickness 3 mm, 110 mAs, 120 kV; 3 minutes per-bed PET) (Siemens, Erlangen, Germany) at the PET/CT unit. Attenuation corrected PET, CT and fusion PET/CT images were reviewed simultaneously; visual and semiquantitative assessments were performed. The maximum standard uptake value (SUV max ) was the Öner et al. 18 18 F-FDG-PET scan. In visual assessment, bone marrow metabolic activity that is greater than the liver was considered to be pathologic. For semiquantitative assessment, regions of interest (ROI) were drawn for SUV max estimation, which included iliac crest in cases with diffusely increased metabolic activity and the highest activity area in cases with focal involvement.

Bone Marrow Biopsy
Unilateral BMB of the posterior iliac crest was performed by different hematologists as part of routine clinical evaluation, and the presence of marrow infiltration was interpreted by an experienced hematopathologist who was blinded to the PET/CT results. Trephine biopsy samples were analyzed following the standard procedures. BMB was considered as positive in the presence of lymphoma involvement. Although flow cytometric immunophenotyping of marrow aspirates can be performed, this method was not used for the diagnosis of bone marrow involvement.

Data Analysis
BM biopsy results were regarded as the reference test for evaluating BMI. Cases with concordant findings in both PET/CT and BMB (both positive or negative) were evaluated as true positive or true negative results. Non-concordance between these two parameters was described as false negativity or false positivity. The sensitivity and specificity rates, positive predictive value (PPV) and negative predictive value (NPV) of PET/CT for detecting BM infiltration were determined for all cases. Additionally, receiver operating characteristics (ROC) curves were formed to determine cut-off values for SUV max . Analyzes were performed with PASW 18 (SPSS/IBM, Chicago, IL, USA) software.

Results
Among the 172 cases, BMI was detected by PET/CT in 33 (19.1%) and by BMB in 42 (24.4%) patients (Table 1). Among the 33 cases with infiltration on PET/CT, 11 had diffuse heterogeneous patchy accumulations while 22 had unifocal/multifocal accumulations. Within the 64 HL patients, BMI was detected by PET/CT in 17 (26.5%) and by BMB in 5 (7.8%) ( Table 2), while among the 108 NHL patients, BMI was detected by PET/CT in 16 (14.8%) and by BMB in 37 (34.2%) ( Table 3). Concordance between PET/CT and BMB was observed in 123 (71%) of 172 patients. Both tests were negative in 110 patients and both were reported positive in 13 patients. A concordance between PET/CT and BMB was detected in 50 (78%) HL patients, both tests were negative in 46 and positive in 4 patients. Non-concordance was observed in 14 (22%) of these patients, 13 patients were positive on PET/ CT but negative on BMB while 1 patient was positive on BMB but negative on PET/CT. Concordance was detected in 73 (67%) NHL patients, both tests were negative in 64 and positive in 9 patients. The tests were non-concordant in 35 patients, 7 patients were positive on PET/CT but negative on BMB while 28 patients were positive on BMB but negative on PET/CT. On visual assessment of all the cases, PET/CT was found to have a 31% sensitivity and 85% specificity rate for detection of BMI with 39% PPV and 79% NPV (Table 4). Visual assessment of HL cases showed 80% sensitivity, 78% specificity with 24% PPV and 98% NPV (Table 4), while in NHL cases the corresponding values were 24%, 90%, 56% and 70%, respectively (Table 4).   (Table 5). Estimated mean SUV max value was 12.02 g/mL for the 33 patients with positive findings on PET/CT; 11.67 g/mL for 13 patients who were positive on both PET/CT and BMB; and 12.25 g/mL for 20 patients who were positive on PET/ CT but negative on BMB (Table 6).

Discussion
Initial evaluation including determination of anatomic distribution of the disease extent is an essential factor to predict both disease-free and overall survival in lymphoma patients. BM involvement in lymphoma indicates generalized disease and is a predictor of poor prognosis.
Besides the role of BMI in primary staging, it has a specific clinical significance for guiding the treatment approach (2). In routine clinical practice, BMB is used to evaluate BM involvement. Although BMB is primarily a safe and risk-free procedure, complications such as bleeding or infection can rarely occur. Additionally, being an invasive and painful procedure can be a disadvantage for patients. In case of insufficient sampling, repeated biopsies may be required. There is no consensus on whether BMB should be performed uni-or bi-laterally, thus biopsies are usually Öner et al. 18 F-FDG-PET/CT Efficiency for Bone Marrow Infiltration Mol Imaging Radionucl Ther 2017;26:69-75    Table 6. Numeric distribution and mean maximum standard uptake values for patients with positive results on positron emission tomography/computerized tomography

PET/CT (+) PET/CT (+)/BMB (+) PET/CT (+)/ BMB (-)
n 33 13 20 Mean SUV max (g/mL) 12 (14), 97 NHL patients were grouped according to the presence of low or high grade disease, and the results of 18 F-FDG-PET/CT scan for initial staging and unilateral iliac crest BMB were compared. Unlike other studies in the literature, samples were obtained from the involvement sites in BMB-negative patients with 18 F-FDG-PET/CT images suggesting BMI. Consequently, BMB from sites of involvement of the 11 patients who were initially BMB-negative and 18 F-FDG-PET/CT positive, revealed 6 positive BMB results. Positive repeat biopsies were obtained from the contralateral iliac crest in 1, from the humerus in 2, from the tibia in 1 and from the fourth vertebra in one patient. Thus, 18 F-FDG-PET/CT was 79% sensitive, 91% specific for detecting BMI with 87% PPV and 87% NPV. Additionally, there was no significant difference between the low and high grade NHL groups in terms of the ability of 18 F-FDG-PET to detect BMI (sensitivity p=0.23, specificity p=0.64). In conclusion, the high potential of 18 F-FDG-PET in detecting BMI in NHL was highlighted and BMB sampling was recommended for BMB negative patients whose 18 F-FDG-PET scan demonstrates BM involvement (14). In our study, biopsies were obtained only from the iliac crest and not from other sites observed positive on PET/CT, which is a limitation of our study. We could have found higher sensitivity and specificity values if biopsy sampling was done from sites other than the iliac crest in cases who were biopsy negative. Many studies in the literature have reported that in the majority of BMBnegative cases multifocal involvement was observed on 18 F-FDG-PET/CT and that biopsies obtained from the sites of involvement were almost always positive (5,15,16,17). In our study, among the 33 cases with infiltration on PET/ CT, 11 had diffuse heterogeneous patchy accumulations. Diffuse accumulations may be secondary to benign conditions such as inflammation, thus some studies (13) have excluded such cases while others have not (11). In our study diffuse accumulations were heterogeneous and patchy, thus they were included since they were not homogenous lesions. In the study by Lee et al. (18), 120 high grade NHL patients comprised of newly diagnosed DLBCL and peripheral T-cell lymphoma cases were included to assess the role of 18  Concordance between the two tests was observed in 73 (67%) patients; 64 patients were PET-CT/BMB negative while 9 patients were PET-CT/BMB positive. Non-concordance was observed in 35 patients; of which 7 were 18 F-FDG-PET/CT positive, BMB negative and the remaining 28 patients were BMB positive, 18 F-FDG-PET/ CT negative. As a result, 18 F-FDG-PET/CT was found to be 24% sensitive, 90% specific for detecting BMI with PPV 56% and NPV 70%. We suggest that the low sensitivity for the NHL subgroup may be due to the lack of histological subgrouping in this patient group. Studies have reported higher 18 F-FDG-PET/CT sensitivity rates in detecting BMI in aggressive NHL subtypes, while 2/3 false negativity ratio was observed in indolent histological forms (e.g., grade 1 and 2 follicular lymphomas) leading to lower sensitivity values (9). Additionally, in our study, AUC and cut-off values for NHL patients were found as 0.6534 and 3.2, respectively.

Conclusion
In this study, a moderately high concordance (71%) was observed between PET/CT and BMB findings. The rate of concordance was higher in HL patients (78%) as compared to NHL patients (67%). In conclusion, PET/CT appears to be a significant method for detecting BM infiltration in comparison to BM biopsy which is an invasive method. Currently, BM biopsy is usually performed from the iliac crest while PET/CT has the advantage of whole body imaging to allow for detection of involvement sites other than the iliac bone.