The usefulness of ¹⁸F-FDG PET/CT for assessing methotrexate-associated lymphoproliferative disorder (MTX-LPD)

This retrospective study was approved by the institutional review board (#013-0422).

We retrospectively analyzed the imaging reports of the patients suspected of having MTX-LPD who underwent PET/CT scanning at Hokkaido University Hospital during the period from June 2009 to December 2014. We searched the electronic database of all PET/CT scans performed during this time period to identify the patients suspected of having MTX-LPD, and the clinical or histological diagnosis was confirmed by review of each patient’s electronic medical records.

A total of 26 patients suspected of having MTX-LPD were identified by the review. The diagnosis of MTX-LPD was made in 22 of these patients. Regarding the other three patients, they were diagnosed with other pathologies (i.e., lung cancer, chronic lymphocytic leukemia, reactive lymph node in one patient each), and the association between MTX treatment and the remaining diagnosed plasma cytoma was unclear; these patients were therefore excluded from the analyses. We also excluded the cases of five patients who were followed up or received additional therapy at another hospital and two patients whose MTX was withdrawn before FDG PET/CT. The final study population was 15 patients with MTX-LPD (62.0 ± 10.7 years old, five men, ten women). Their characteristics are summarized in Table 1. The reasons for MTX therapy were RA (n = 13), polyarteritis nodosa (n = 1), and psoriatic arthritis (n = 1). The mean duration of MTX treatment was 84.1 ± 59.6 months (range 7–234 months). Nodal regions of lymphoma were observed in 11 of the 15 patients. Extranodal regions of lymphoma were observed in eight of the patients, including the gingiva (n = 2), skin/subcutaneous site (n = 4), lung (n = 4), liver (n = 2), pericardium (n = 1), bowel (n = 1), bone (n = 1), tonsil (n = 1) and adrenal (n = 1).

Pathological data were obtained using tissue specimens obtained by biopsy or resection from 11 patients. The other four patients were diagnosed based on their clinical course. The pathologically confirmed histological types were diffuse large B-cell lymphoma (n = 5), pleomorphic B-cell lymphoproliferative disease (n = 3), follicular lymphoma (n = 1), mucosa-associated lymphoid tissue lymphoma (n = 1), and marginal zone lymphoma (n = 1). After the withdrawal of MTX, nine of the 15 patients (60.0 %) achieved a CR, and the other six patients (40.0 %) were non-CR. The mean follow-up duration was 31.7 ± 19.4 months (range 9–67 months).

The histopathological diagnosis was determined on the basis of the WHO lymphoma classification criteria [2] for the 11 patients whose pathological data were available. In addition to hematoxylin-eosin staining, immunohistochemical staining, flow cytometry analysis, and an EBV in situ hybridization analysis were performed for pathologic confirmation. The follow-up evaluation was established on the basis of the clinical follow-up, which included clinical examinations, ultrasonography, contrast-enhanced CT, ¹⁸F-FDG PET /CT scanning, or an additional biopsy every 1–3 months.

In this study, the patient outcome was defined as a CR when all lesions of MTX-LPD had disappeared after the withdrawal of MTX without cytotoxic treatment (i.e., chemotherapy and radiotherapy) during the follow-up. The patient outcome was defined as non-CR when additional chemotherapy was needed to treat the residual tumor after the withdrawal of MTX or the patient had residual tumor or a recurrence during the follow-up period. We defined the clinical outcome at the time of our review.

All patients underwent ¹⁸F-FDG PET/CT acquisitions using an integrated PET/CT scanner (Biograph 64 PET/CT scanner, Asahi-Siemens Medical Technologies, Tokyo). Before tracer injection, the patient fasted for at least 6 h. Following a blood glucose test to confirm blood glucose levels less than 150 mg/dL, PET images were acquired 60 min after an intravenous injection of ¹⁸F-FDG (4–5 MBq/kg). Emission scanning for 3 min per bed was carried out following the CT image acquisition for attenuation corrections.

The acquired datasets were corrected for attenuation, dead-time and scatter, and the images were reconstructed using a point spread function-based iterative algorithm (TrueX, Siemens) [18] with two iterations per 21 subsets with 512 × 512-pixel matrix, a matrix size of 168 × 168, a voxel size of 4.1 × 4.1 × 2.0 mm, and a Gaussian filter at 4.0- mm full-width at half-maximum. The transaxial and axial field of views were 58.5 cm and 21.6 cm, respectively.

All patients were scanned with either a 64-slice or a 320-slice multi-detector row CT scanner. Six patients underwent a contrast-enhanced scan. These patients received an intravenous injection of iodine contrast agent (450–560 mgI/kg) at a rate of 2–3 mL/s using a power injector, and CT images were obtained approx. 60 s following the injection. The other nine patients underwent a non-contrast enhanced MDCT scan. The scan range of 12 patients was from the middle of the skull to the mid-thigh. The scan range of the other patients was the cervical region only, the abdominal region only, or from the cervical to chest region. Axial CT images were reconstructed using a 5-mm section thickness.

Each patient’s disease stage was determined by the Ann Arbor classification system for malignant lymphoma [19]. Staging was confirmed by clinical follow-up or biopsy. We also divided the whole body into 16 nodal regions and nine extranodal anatomic regions for analysis, in accord with previous studies [20‐22] (Table 2).

We compared each patient’s FDG-PET/CT findings with the results of the corresponding imaging examinations in terms of the presence of MTX-LPD disease. We classified the findings as follows. TP: true-positive (presence of MTX-LPD), TN: true-negative (absence of MTX-LPD), FP: false-positive (abnormal FDG uptake unrelated to MTX-LPD), or FN: false-negative (missed diagnosis of proved MTX-LPD) according to the reference standard. The reference standard included histopathologic findings (obtained by biopsy) or informative follow-up (clinical, laboratory, PET/CT, or other imaging findings such as endoscopy, MRI, and ultrasonography). It was impossible to obtain histopathologic proof of all of the suspected lesions in most of the clinical situations, especially in the patients with systemic disease spread. We therefore set the reference standard by using the method comparing PET/CT and MDCT findings in studies similar to ours [23, 24].

For example, in the case of a patient entering remission, lesions that were resolved on follow-up imaging were considered TP, and lesions that remained stable or progressed were considered FP. In the case of a patient experiencing disease progression, lesions that progressed were considered TP, and lesions that resolved or remained stable on follow-up imaging were considered FP. A region was considered FN if a lesion was not seen by one modality but was identified by another modality and met the criteria for a TP result as described above. A region was considered an FN if lesions were seen but were considered non-malignant on FDG-PET/CT or MDCT and disease developed at the same site of follow-up. Regions were considered TN when the site remained disease-free at the follow-up after MTX withdrawal.

The clinical images were independently evaluated in random order by two nuclear medicine physicians (S.W. and O.M.) for ¹⁸F-FDG PET/CT and two diagnostic radiologists (Y.K. and N.M.) for MDCT. The readers were blinded to all clinical information, the results of the other imaging modalities, such as endoscopy, MRI, and ultrasonography before MTX withdrawal, and the evaluation by the other reader.

Thirty-three lymph node regions and 18 extranodal anatomic regions were out of the field of view of the anatomical CT scanning. Therefore, a final total of 207 lymph node regions and 117 extranodal anatomic regions were analyzed. Areas with focally increased ¹⁸F-FDG uptake were considered to be sites of active disease for PET/CT. For the MDCT analysis, abnormal lymph nodes were defined as having a minimum dia. >1 cm and extranodal lesion as mass lesions considered as equivocal or positive for malignancy. All regions were evaluated as positive or negative. If there was a discordant finding between the two readers, consensus was obtained by discussion. Lesions considered positive, suggestive or equivocal for MTX-LPD were considered positive for the analysis. Lesions reported as unlikely or negative for MTX-LPD were considered negative for the analysis. The number of disease sites was evaluated by each physician.

In addition, one nuclear medicine physician (S.W.) assessed the ¹⁸F-FDG PET images using semi-quantitative methods. The SUVmax from the single pixel showing the highest FDG accumulation in the lesion was calculated to obtain the information of tumor activity, as [tissue radioactivity concentration (Bq/ml)] × [body weight (g)] / [injected radioactivity (Bq)]. Two volume-based parameters, the WBMTV and the WBTLG, were also measured.

The WBMTV was calculated from the ¹⁸F-FDG PET images according to the following procedure. First, the boundaries of voxels with an SUV intensity exceeding 2.5 were produced automatically. Second, the physiological uptake including that in the brain, oral cavity, pharynx, heart, stomach, liver, intestines, kidney, ureter, bladder, skeletal muscle, and any other tissues was carefully subtracted by the nuclear medicine physician. Third, false-positive lesions, such as inflammation of joints or other benign ¹⁸F-FDG-avid lesions based on MDCT or follow-up studies including physiological examination, MDCT, ultrasonography and PET/CT scan were subtracted. Finally, the total lesion glycolysis (TLG) value was calculated for every target lesion as TLG = MTV × SUVmean, and the WBTLG was calculated as the total of TLG in the entire body.

Data are expressed as the median and range. Differences in the SUVmax, WBMTV and WBTLG between two groups were tested using Welch’s t-test for categorical variables. A Bowker test was used to compare the accuracy of ¹⁸F-FDG PET/CT scans and MDCTs in detecting malignant lesions. For each analysis, p-values <0.05 were considered significant. Statistical calculations were carried out using statistical software (JMP ver. 10, SAS, Cary, NC, USA).

MTX-LPD lesions were analyzed in 207 nodal regions and 117 extranodal regions and identified in 75 nodal anatomic regions (36.2 %) and 17 extranodal anatomic regions (14.5 %). ¹⁸F-FDG PET/CT was superior to MDCT at detecting both nodal and extranodal regions. ¹⁸F-FDG PET/CT scanning could detect 70/75 nodal (93.3 %) and 11/17 extranodal anatomic regions (64.7 %), whereas MDCT detected only 45/75 nodal (60.0 %) and 9/17 extranodal anatomic regions (52.9 %). Consequently, the sensitivity, specificity, and accuracy (Table 3.) for the detection of lesions were 83/92 (90.2 %), 226/232 (97.4 %), and 309/324 (95.4 %), respectively, for ¹⁸F-FDG PET/CT and 55/92 (59.8 %), 220/232 (94.8 %), and 275/324 (84.9 %), respectively, for MDCT (Table 4). The diagnostic accuracy of FDG-PET/CT was significantly higher than that of MDCT (p < 0.002). In the evaluation of interobserver agreement, discordant assessments were observed for five nodal regions and none of the extranodal regions in the FDG PET/CT reading, and for 30 nodal regions and five extranodal regions in the MDCT reading.

There were five extranodal lesions that ¹⁸F-FDG PET/CT scanning could identify but MDCT could not; one in the bowel, one in bone (Fig. 1), one in the tonsil and two in the gingiva. The FN lesions were five nodal regions; both sides of the infraclavicular, the right axilla, and both sides of the inguinal and femoral regions. Four extranodal regions were FN; three in skin detected by only visual examination and one in lung. There were six FP lesions; lung, bone and four nodal lesions in the axilla, hilar, mesenteric, and iliac regions. One was a lung lesion that showed focal ¹⁸F-FDG uptake; this was pneumonitis. A second FP lesion was on bone that had focal ¹⁸F-FDG uptake, which was shown to be negative by bone marrow aspiration cytology. The other FPs identified by ¹⁸F-FDG PET/CT scanning were four nodal lesions that include the axilla, hilar, mesenteric and iliac regions with asymmetric ¹⁸F-FDG uptake.

The medians and ranges of the ¹⁸F-FDG PET parameters were as follows. SUVmax: 10.3 (range 0–47.1), WBMTV: 33.5 ml (range 0–362.6 ml), and WBTLG: 167.0 ml (range 0–2180.9 ml). The differences in the metabolic parameters between the CR patients and the non-CR patients are presented in Table 5. There was no significant association between ¹⁸F-FDG PET/CT parameters and the CR patients.