Introduction
Lymphoma is the third most common malignancy in children and adolescents. About 60% account for non-Hodgkin lymphoma (NHL) and 40% for Hodgkin lymphoma (HL). In the past 40 years, major advances were achieved in the fields of cytotoxic drug development, radiation oncology, pathology and imaging technologies. Each of them contributed to a more individualized treatment.
The role of the latest whole-body imaging technologies, [
18F]-FDG-PET/CT and [
18F]-FDG-PET/MRI, has been studied, but much more extensively in pediatric HL than in pediatric NHL [
1]. Yet, according to the PubMed database, to date there is no publication available which exclusively addresses the application of simultaneous [
18F]-FDG-PET/MRI in pediatric NHL (last PubMed query: April, 14, 2021).
In 2015, the revised International Pediatric non-Hodgkin Lymphoma Staging System (IPNHLSS) was published and since then replaces the St. Jude classification by Murphy from the 1980s [
2]. The main publication on the revised IPNHLSS also considers advances in imaging technologies and points out the efficiency and potency of hybrid imaging such as [
18F]-FDG-PET/CT and [
18F]-FDG-PET/MRI [
2]. [
18F]-FDG-PET/MRI seems to be especially promising since it can depict metabolic (i.e., the degree of glucose turnover), functional (i.e., the diffusion movement of water molecules) and morphological properties of tumor lesions at the same time. Considering all three aspects combined could be substantial for accurate characterization of lesions and may reduce the deficiencies of each modality when used separately. Furthermore, radiation exposure can be saved when replacing CT imaging by MRI.
However, whole-body imaging with [
18F]-FDG-PET/MRI represents a compromise between the acquisition of all potentially possible MR sequences and a reasonable overall scan duration [
3]. The overall scan duration is the determining factor with regard to compliance of young patients. It should not exceed 60 min [
3].
The aim of this retrospective single-center evaluation was to investigate the performance of whole-body [18F]-FDG-PET/MRI in pediatric NHL patients by using a limited number of MRI sequences.
Patients, materials and methods
Patients
Between April 2012 and November 2019 ten pediatric NHL patients received whole-body [18F]-FDG-PET/MRI for initial staging.
Written informed consent was obtained from all patients and/or their legal guardians for scientific evaluation of imaging data before inclusion into this retrospective evaluation.
The data evaluation was approved by the local Ethics Committee.
Patient data and investigations beyond [18F]-FDG-PET/MRI
For each patient, age, sex, histology, the presence of B symptoms and the level of lactate dehydrogenase (LDH) were documented. Wherever available, the results of ear, nose and throat (ENT) examination, lumbar puncture, bone marrow (BM) biopsy, thoracentesis as well as computed tomography (CT) of the chest and ultrasound of the abdomen were considered for correlation with the [18F]-FDG-PET/MRI images.
[18F]-FDG-PET/MRI imaging
Images were acquired on a 3 Tesla Biograph mMR (Siemens, Erlangen, Germany) which had been installed in 2011.
Prior to the [18F]-FDG injection all patients fasted for at least 6 h.
To avoid activation of brown adipose tissue all patients were warmed for at least 30 min before [18F]-FDG injection. In addition, the unselective beta-blocker propranolol (1 mg/kg, maximum dose = 40 mg) was administered approximately one hour before tracer injection.
The upper limit of [
18F]-FDG activity to be administered was determined by the EANM dosage calculator [
4]. Dosing reduction ranged from 24 to 67% of the recommended upper activity limit in nine patients (Table
1). In one patient with marginal overweight the administered activity reached the recommended upper limit (Table
1).
Table 1
Patient characteristics [18F]-FDG-PET/MRI parameters and results of further staging examinations
1 | 15 years | Burkitt | n.d | n.d | − | − | Splenomegaly, otherwise US - | 212 MBq (− 39%) | 348 MBq | 90 min | No | WB PET; WB MR T2-TIRMcor + trans; WB DWItrans | Nodular lymphoma involvement on both sides of the diaphragm | |
2 | 16 years | DLBCLoss | − | − | − | + | − | 276 MBq (− 24%) | 363 MBq | 90 min | No | WB PET; WB MR T2-TIRMcor + trans; WB DWItrans | Nodular lymphoma involvement on both sides of the diaphragm, extensive involvement of the left pelvic bones with cortical bone destruction and infiltration of adjacent muscles | |
3 | 6 years | DLBCL | − | n.d | − | − | − | 64 MBq (− 49%) | 126 MBq | 135 min | Yes | WB PET; WB MR T2-TIRMcor + trans; WB DWItrans | Nodular lymphoma involvement on both sides of the diaphragm | |
4 | 7 years | DLBCL | − | n.d | − | − | − | 91 MBq (− 39%) | 148 MBq | 110 min | No | WB PET; WB MR T2-TIRMcor + trans; WB DWItrans | Nodular lymphoma involvement on both sides of the diaphragm | |
5 | 10 years | DLBCLoss | − | n.d | − | − | − | 126 MBq (− 45%) | 229 MBq | 180 min | No | WB PET; WB MR T2-TIRMcor + trans; WB DWItrans | Long segmental lymphoma involvement of the left femur with cortical bone destruction and infiltration of adjacent muscles | |
6 | 10 years | T-cell lymphoma | − | n.d | − | − | Bilateral renal involvement, otherwise US - | 278 MBq (+ 0%) | 277 MBq | 110 min | No | WB PET; WB MR T2-TIRMcor + trans; WB DWItrans | Nodular lymphoma involvement on both sides of the diaphragm including Waldeyer’s ring. In addition bilateral kidney and skeletal involvement (right femur) | |
7 | 14 years | Burkitt | − | n.d | − | − | − | 162 MBq (− 45%) | 292 MBq | 45 min | No | WB PET; WB MR T2-TIRMcor + trans; WB DWItrans | Nodular lymphoma involvement above diaphragm including Waldeyer’s ring. Moreover, focal skeletal involvement | |
8 | 15 years | T-cell lymphoma | + | n.d | − | − | Bilateral renal involvement, otherwise US - | 67 MBq (− 67%) | 200 MBq | 70 min | No | WB PET; WB MR T2-TIRMcor + trans; WB DWItrans | Nodular lymphoma involvement above diaphragm including left pleura, extended bilateral and focal skeletal involvement (left femur). Waldeyer’s ring involvement most unlikely | |
9 | 4 years | T-cell lymphoma | + | + | − | − | − | 45 MBq (− 51%) | 92 MBq | 70 min | No | WB PET; WB MR T2-TIRMcor + trans; WB DWItrans n.e | Nodular lymphoma involvement above diaphragm and left-sided pleural involvement. Waldyer’s ring involvement most unlikely | |
10 | 17 years | DLBCL | − | n.d | − | − | − | 239 MBq (− 31%) | 348 MBq | 80 min | No | WB PET; WB MR T2-TIRMcor + trans WB DWItrans n. d | Nodular lymphoma involvement above diaphragm | |
Time interval between radiotracer injection and start of the scan ranged from 45 to 135 min for nine patients (Table
1). The variation in injection to imaging time is mainly due to unexpected demand on the scanner throughout daily routine. In one patient the interval was 180 min due to temporary scanner dysfunction (Table
1).
One of the ten patients needed sedation for [
18F]-FDG-PET/MRI acquisition (Table
1).
[
18F]-FDG attenuation correction was carried out per section with a MR Dixon sequence as described by Hirsch et al. [
3].
Water-sensitive fast inversion recovery sequences (T2-TIRM) with 5 mm in coronal and 4.5 mm in transversal plane were acquired to provide anatomical coverage of the entire body. This was limited to a maximum length of 160 cm (seven stacks with overlap of 30 cm length each). Taller children and adolescents underwent a second scan covering the legs after repositioning.
Respiratory triggering with a belt system was used for image acquisition of the thoracic and upper abdominal region in order to obtain improved image quality and to minimize breathing artifacts.
To reduce artifacts through excessive peristalsis all patients received butylscopolamine approximately 10–15 min before acquisition of the abdominal region was performed (0.3 mg/kg, maximum dose = 20 mg).
Additionally, diffusion-weighted images (DWI) with a slice thickness of 6 mm were acquired if patient compliance was sufficient. This was the case in eight of the ten patients (Table
1). One patient, however, refused an additional DWI whereas a second patient was unable to lie still during DWI acquisition resulting in severe movement artifacts. DWI allowed for calculation of the apparent diffusion coefficient (ADC). ADC cutoff values (B-values) at or less than 800 × 10
–6 mm
2/s were regarded as diffusion restriction.
Altogether, the acquisition with its detailed parameters was in accordance with recommendations by Hirsch et al. [
3].
[18F]-FDG-PET/MRI data analysis
[
18F]-FDG-PET/MRI datasets were retrospectively analyzed by an experienced pediatric radiologist (CR) and an experienced nuclear medicine physician (LK), both having more than 10 years’ experience in their respective field. Nineteen lymph node regions, adapted to [
5], two extranodal regions [Waldeyer’s ring (WR), pleura] as well as six organs [central nervous system (CNS), lungs, liver, spleen, kidneys and skeleton) were defined for evaluation.
First, [18F]-FDG-PET images were evaluated by the nuclear medicine physician and MR images (T2-TIRM coronal and transversal) by the pediatric radiologist. The nuclear medicine physician focused on the aspects of glucose metabolism (intensity and configuration of uptake) whereas the radiologist reviewed aspects of morphology (diameters and intensity of the T2 signals). Thereby, the above-mentioned regions were classified as (a) involved = positive, (b) not involved = negative. The results were documented in separate SPSS tables (SPSS24).
In a second step, both physicians reviewed discrepant findings jointly. For this purpose, diffusion weighted MR images were also reviewed whenever available. Consensus was reached by weighing the respective image information ([18F]-FDG, T2-TIRM coronal and transversal, diffusion-weighted images) and discussing them with other experienced colleagues.
Third, findings on [18F]-FDG-PET/MR images were correlated with the results of other clinical investigations or imaging modalities as mentioned above.
Discussion
A new staging system for pediatric NHL was released in 2015 [
2]. The respective publication also mentions the developments of cross-sectional imaging over the past decades including the latest which is [
18F]-FDG-PET/MRI [
2]. We performed a single-center evaluation with ten pediatric NHL patients at the time of initial diagnosis in order to investigate the performance of [
18F]-FDG-PET/MRI by using a limited number of MRI sequences. A focus was put on lymph node regions as well as organs and several extranodal sites of potential lymphoma involvement.
To date publications evaluating the application of [
18F]-FDG-PET and MRI (including DWI) in NHL patients stem from imaging procedures performed separately of each other. Thereby, time intervals of up to two weeks are described [
7‐
9]. However, during such time intervals lymphoma may have further developed and spread. Furthermore, other non-malignant diseases like tonsillitis or pneumonia may occur or resolve in the meantime and could lead to differing results. In this respect, a comparison is not valid. In contrast to that, simultaneous [
18F]-FDG-PET/MRI allows imaging without any delay between the two modalities and therefore a more reliable comparison of the respective findings.
In our evaluation, a total of 190 lymph node regions were assessed. For 186 of 190 (98%), [
18F]-FDG-PET and MRI came to concordant results. This is in line with the literature describing high [
18F]-FDG-PET avidity for most lymphoma subtypes and excellent lymph node accessibility with whole-body MRI [
10‐
12]. For the few cases with discrepant findings the addition of diffusion weighted images (DWI) facilitated decision making. Using the cutoff value at 800 × 10
–6 mm
2/s for non-Hodgkin lymphoma, which was also suggested by Kwee et al. [
13] and Vermoolen et al. [
14], resulted in a clear differentiation between likely and unlikely lymphoma involvement. We were aware that motion artifacts, which often occur in the areas of neck, chest and abdomen, may impair the reliability of DWI and ADC values [
15]. In our limited cohort, however, such problems were not noticed which may be due to the increased efforts in handling pediatric patients (e.g., sufficient immobilization during [
18F]-FDG-PET/MRI acquisition or the use of respiratory triggering) [
3]. Thus, further evaluation of the potential capacity of DWI seems warranted in order to establish this functional information as a third cornerstone beside morphology and metabolism.
The Waldeyer’s ring (WR) turned out to be a region with noticeable discrepancies between clinical examination and [
18F]-FDG-PET/MRI: In two patients, suspicion of WR involvement based on ENT exam could not be confirmed on [
18F]-FDG-PET/MR images. On the other hand, there were two patients showing highly suspicious pattern of their WR on [
18F]-FDG-PET/MRI with unremarkable ENT exam. Those discrepancies can result from different facts: First, ENT exam as an inspection of the oral cavity, regularly performed by pediatric oncologists, can only evaluate the oro-pharynx. The nasopharynx (e.g., pharyngeal tonsil) remains unevaluated if not examined by an ENT specialist using dedicated instruments. Second, even if the entire WR is examined by an ENT specialist, only the mucosal surface is evaluated but not the underlying tissue. Furthermore, alterations of the mucosa can be unspecific and therefore difficult to interpret. This makes ENT examinations highly subjective and therefore prone to interobserver variability [
6,
16]. Thus, an evaluation of the WR should always include cross-sectional imaging, preferably [
18F]-FDG-PET/MRI. The detection of WR involvement is crucial to anticipate airway obstruction early on especially in patients with NHL entities often affecting the WR such as Burkitt lymphoma and DLBCL [
16,
17]. Furthermore, there is evidence that NHL patients with WR involvement may benefit from irradiation of their WR [
18,
19].
Concerning pleura involvement, the results of our study support previous assumptions that [
18F]-FDG-PET can differentiate between pleural effusion and malignant pleura involvement [
20]. Pleural assessment, however, is challenging on whole-body MR sequences, especially if solely coronal slices are acquired [
21,
22]: Due to the thoracic convexity, coronal slices are prone to partial volume effects. Partial volume effects considerable complicate the assessment of the pleura, so that small lesions in particular are easily missed. On transversal slices, by contrast, partial volume effects are less frequent, making them more suitable for pleural assessment. Another option would be a dedicated lung MRI. However, this would significantly prolong imaging time and contradicts the intention of whole-body MRI which is to provide a fast overview without loss of compliance in young patients due to long imaging time.
Splenic involvement occurs in about 30–40% of all NHL patients [
23]. However, in the cohort analyzed here, none of the patients had suspicious focal lesions within splenic parenchyma, neither on ultrasound nor on MRI or [
18F]-FDG-PET. Nevertheless, in one patient splenomegaly was detectable. Splenomegaly is challenging because it is a relatively unspecific sign. It can accompany spleen involvement as well as various pathologies like infectious diseases (e.g., mononucleosis, malaria), metabolic disorders (e.g., Morbus Gaucher) or liver diseases resulting in portal hypertension [
24]. In case of normal parenchymal texture some publications suggest comparing radiotracer uptake of the spleen with liver and/or bone marrow uptake. If tracer uptake of the spleen exceeds tracer uptake of liver and/or bone marrow, it is recommended to assume spleen involvement [
25,
26]. In our patient with splenomegaly on ultrasound, radiotracer uptake of the spleen did not exceed that of the liver or the bone marrow.
Kidney involvement was detected in two of the ten patients. Both patients had bilateral renomegaly due to extensive lymphoma infiltration. In young NHL patients with kidney involvement, multiple bilateral large lesions are the most commonly found pattern [
27,
28]. Both MRI (including DWI) and [
18F]-FDG-PET were able to correctly detect kidney involvement. This is important to mention for [
18F]-FDG-PET, since kidney evaluation with [
18F]-FDG-PET is often regarded as extremely challenging due to renal excretion of the radiotracer. However, cortical lesions and manifestations distant from the collecting system are well distinguishable from physiological tracer excretion as it was in our two patients.
Skeletal involvement in NHL patients can be either focal or diffuse. Literature review suggests a pattern of rather focal involvement in patients with DLBCL, whereas other NHL entities tend toward diffuse bone marrow infiltration [
29‐
31].
Bone marrow biopsy is much more sensitive in detecting diffuse BM infiltration compared to [
18F]-FDG-PET and MRI [
32‐
35]. In contrast, a focal involvement pattern is detectable on [18F]-FDG-PET and MR images with the highest sensitivity [
32‐
34]. All six patients from our cohort who had skeletal involvement (only one of them had DLBCL) exhibited a pattern of focal involvement which was well detectable through [
18F]-FDG-PET/MRI. However, BM biopsy taken from the iliac crest was negative in five out of the six cases. It is worth mentioning that BM biopsy was also negative in the patient in whom one of the BM lesions was not far located from the needle track. This suggests that BM biopsy probably only yields a positive result if the biopsy needle hits the lesion visible on imaging, which, in turn, may explain the low sensitivity of bone marrow biopsies in case of a focal involvement pattern.
In one of the six patients [
18F]-FDG-PET/MRI and BM biopsy were positive. This constellation could be of prognostic value: Based on an analysis of 327 adult DLBCL patients, Cerci et al. [
36] found that patients who were positive in both modalities, [
18F]-FDG-PET and BM biopsy, had a worse outcome compared to patients who were positive in only one modality. However, in our patient the needle track ran directly through the lymphoma manifestation. Thus, in this patient positivity of both modalities does not necessarily imply a worse outcome.
Several limitations of this retrospective evaluation need to be mentioned:
First, the number of cases is small. Thus, the described results should be confirmed, preferably within a large multicenter trial with predefined image acquisition parameters.
Second, the investigated cohort does not cover the typical spectrum of childhood NHL entities, which are Burkitt lymphoma, lymphoblastic lymphoma and anaplastic large-cell lymphoma. The spectrum of this study refers to an adolescent population in which DLBCL is more common.
Third, our study population does not seem to include high-risk patients. Particularly LDH, as one of the main risk factors, was relatively low in all cases.
Forth, some of the potential sites of extranodal NHL involvement like CNS, liver and lung were not evident in the analyzed cohort, precluding the assessment of the full potential of [18F]-FDG-PET/MR imaging.
Fifth, despite being able to show that the applied DWI performed excellent, it has to be mentioned that new developments concerning DWI have evolved, e.g., whole-body-DWI with background suppression (DWIBS) [
37]. The latter provides superior local resolution and reduced artifact susceptibility [
38]. It is characterized by a relatively short acquisition time, which is particularly desirable in the setting of whole-body imaging in children and adolescents [
39]. And a robust fat suppression has a positive effect on subsequent three-dimensional reconstruction. In young children, however, the physiological diffusion restriction of the pelvis and spine has to be taken into account, since it could lead to false positive findings [
39]. New opportunities to display DWI findings (e.g., MIP grey-scale inverted DWI) might also facilitate direct comparison between function and metabolism.
Sixth, the clinical and paraclinical results were used as standard of reference to some of the imaging results. However, clinical and paraclinical parameters have limitations as well. To overcome these limitations, confirmation through biopsy would be necessary. The latter, however, is often not feasible nor justifiable in the clinical setting as well as ethically.
Seventh, the aim of this study was to evaluate the performance of whole-body [18F]-FDG-PET/MRI for a limited overall acquisition time. It was not the aim to compare [18F]-FDG-PET and MRI. For that, a broader spectrum of MRI sequences would have been necessary leading to a considerable prolongation of image acquisition time, which is difficult in a pediatric age group. However, imaging technology is developing quickly. Thus, in the future a broader spectrum of MR sequences can be expected to be acquired within one hour of acquisition time.