Introduction

Fluorine-18 fluorodeoxyglucose (FDG)-positron emission tomography (PET) has been shown to improve the accuracy of lymphoma staging for Hodgkin lymphoma and B-cell lymphomas [1, 2]. Accordingly, the International Working Group has also incorporated PET or PET-computerized tomography (CT) in the assessment of treatment responses for Hodgkin lymphoma and diffuse large B-cell lymphoma [3, 4]. Emerging data in Hodgkin lymphoma also suggest that PET is prognostically relevant [5], and may even surpass the International Prognostic Score in predicting survival [6].

However, data on the value of PET in mature T-cell and natural killer (NK)-cell lymphomas are scarce, owing to their lower incidence, especially in Caucasian populations [7]. In this study, we analysed the PET findings of a consecutive cohort of patients with mature T-cell and NK-cell lymphomas, in order to define the role of PET in staging, definition of functional tumour load and the correlation with clinicopathological features and treatment outcome.

Materials and methods

Patients

In the year 2007, consecutive patients with mature T-cell and NK-cell lymphomas, who required staging for newly diagnosed disease, or radiological re-assessment for relapsed or refractory diseases, were studied. The lymphoma diagnosis was based on the World Health Organization lymphoma classification [8].

Staging investigations

Patients underwent standard investigations for staging, including bilateral marrow trephine biopsies. Testing for Epstein–Barr virus (EBV) in the lymphoma cells was performed by in situ hybridisation (ISH) for EBV encoded small RNA (EBER) [9]. Circulating plasma EBV DNA in patients with EBV-positive lymphomas, a surrogate biomarker for tumour load, was quantified by real-time quantitative polymerase chain reaction as described [10].

PET-CT scan

Whole body FDG-PET-CT was performed (Discovery VCT; 64-MSCT, GE Healthcare Bio-Sciences Corp., Piscataway, NJ, USA; field of view, 50 cm; pixel size, 3.91 mm; spiral CT pitch, 0.984:1; gantry rotation speed, 0.5 s; FDG dose, 222–370 MBq; uptake, 60 min). A whole body emission PET scan was obtained with five bed positions within 20 min. Attenuation-corrected PET images with CT data were reconstructed with an ordered-subset expectation maximisation iterative reconstruction algorithm (14 subsets and two iterations). The CT images were reconstructed at 2.5-mm intervals to fuse with the PET images (Advanced Workstation 4.3, GE Healthcare Bio-Sciences). FDG uptakes above background or the liver/spleen were considered abnormal, and the highest standardised uptake value (SUVmax, corrected for body weight) was recorded. On CT, lymph nodes above 1 cm (or 1.5 cm in the groin) in short-axis diameter were considered positive. Reporting of the PET-CT scan was conducted without prior knowledge of the results of the other staging investigations.

Statistical analyses

Survival and remission status were censored on the last day of 2007. Statistical analyses (Mann–Whitney U-test, Fischer’s exact test and correlation test) were performed on SPSS11.0 (Chicago, IL, USA). Owing to limitation of case number, data on actuarial survival and multivariate analyses were not obtained.

Results

Patients

There were 24 men and six women, at a median age of 48 (17–79) years (Table 1). The diagnoses included extranodal NK-cell lymphoma, nasal type (n = 12; nasal, 10; extranasal, 2), angioimmunoblastic T-cell lymphoma (AILT, n = 7), anaplastic large cell lymphoma (ALCL, n = 4; systemic, 3; cutaneous, 1), peripheral T-cell lymphoma, unspecified (PTCL-U, n = 4), mycosis fungoides (MF, n = 2) and T-cell large granular lymphocytic (T-LGL) leukaemia (n = 1). Nineteen patients were investigated at diagnosis. Five patients were investigated at relapse (median, 14; 10–124 months from diagnosis), and six at disease progression (median, 4; 1–9 months from diagnosis). For all relapsed or refractory patients, at least 3 weeks had elapsed between systemic chemotherapy and PET.

Table 1 Clinicopathological and PET features of 30 cases of mature T-cell and natural killer cell lymphomas
Full size table

Investigation results

The median lactate dehydrogenase (LDH) level at PET-CT for all cases was 466 (230–2,343; reference range, 197–401) IU/ml, being normal in nine patients. For 12 cases of NK-cell lymphoma, all of which were EBV-positive, the median circulating EBV-DNA level was 1.9 × 104 (<102 to 2.5 × 106) copies/ml.

NK-cell lymphomas

All 12 NK-cell lymphomas were intensely FDG-avid (mean SUVmax, 8, 6.9–10.0) in both nasal (10 patients) and extra-nasal sites (five patients, in bone marrow, bowel, adrenal glands, eye, pericardium, testis and lymph nodes). In 10 cases, the PET and CT were comparable in defining the sites of disease involvement, giving results that were consistent with clinicopathological findings and staging (Fig. 1a,b). On the other hand, for the assessment of local disease extent in the nasal area and maxillary sinuses in all the cases of nasal NK-cell lymphoma, the FDG-avid tumour was much more localised than that shown on CT, suggesting that the soft tissue masses evident on CT were due to inflammation (Fig. 1c). In the two patients with disseminated lymphoma that involved the gut and testis, there was no abnormal marrow FDG uptake. However, occult marrow infiltration was uncovered by positive EBER-ISH [11], although neoplastic infiltration could not be discerned on morphological examination.

Fig. 1
figure 1

Natural killer cell lymphoma. a Positron emission tomography-computerized tomography showing testicular NK-cell lymphoma. This patient had involvement of the left eyelid, left nasal cavity, left maxillary sinus, pericardium, small bowel and right testis. b PET showing NK-cell lymphoma involving the pericardium and bowel, with corresponding findings on fused PET-CT (arrows). c Recurrent nasal NK-cell lymphoma. Plain CT showed obliteration of the nasal cavity, bilateral ethmoid and sphenoid sinuses by soft tissue densities. Biopsies of most of these sites were negative for lymphoma cells. Fused PET-CT showed a hypermetabolic focus at the superior aspect of the left nasal cavity, which indicated the active tumour component

Full size image

Nodal and systemic mature T-cell lymphomas

All fourteen cases of AILT, PTCL-U and systemic ALCL showed avid FDG uptake (Fig. 2). In general, CT and/or biopsy-documented sites of involvement (lung, bone marrow, gut and lymph node) were concordantly PET-positive. However, in some cases individual sites might show discordant findings. In five cases (AILT, n = 2; ALCL, n = 2; PTCL-U, n = 1), lymph nodes considered insignificant (<1 cm) on CT were FDG-avid, suggesting disease involvement. Furthermore, one case of PTCL-U showed marrow infiltration, which was not discernable on CT or by bilateral trephine biopsies (Fig. 3a,b). This was likely because FDG-avid marrow areas were distant from the posterior superior iliac crests that were the sites of trephine biopsies. Finally, in the only case of T-LGL leukaemia investigated, despite the presence of circulating T-LGL leukaemic cells, progressive lymphadenopathy and hepatomegaly evident on CT, and biopsies showing infiltration of lymph nodes, liver and bowel, there was no significant increase in FDG uptake in any anatomical sites (Fig. 4a,b).

Fig. 2
figure 2

Relapsed angioimmunoblastic T-cell lymphoma. Positron emission tomography showed stage IV disease with nodal involvement in the cervical, mediastinal, axillary, hilar, mesenteric, para-aortic and iliac groups (arrows); diffuse marrow and also splenic involvement (thick arrow). The extent of disease was much better delineated than organomegaly shown on computerized tomography

Full size image
Fig. 3
figure 3

Peripheral T-cell lymphoma—unspecified with focal marrow fluorodeoxyglucose uptake. Fused positron emission tomography-computerized tomography of the second cervical vertebra (a) and bilateral acetabuli (b), showing focal FDG uptake (arrows) in the spinous process and left acetabulum respectively. Bilateral posterior superior iliac crest trephine marrow biopsies were negative for lymphoma infiltration

Full size image
Fig. 4
figure 4

T-cell large granular lymphocyte leukaemia. a Computerized tomography showing progressive mesenteric lymphadenopathy (arrows) in terms of size and number, and bowel wall thickening, with disease documented by biopsy in gut, liver, marrow and blood. b Fused positron emission tomography-computerized tomography showing no significant fluorodeoxyglucose uptake

Full size image

Cutaneous mature T-cell lymphoma

In two cases of MF, the skin lesions and the involved lymph nodes in one case showed low FDG uptake. PET results were concordant with findings on clinical examination and CT scanning. One case of cutaneous ALCL also showed minimal uptake, consistent with the spontaneous regression that occurred during the staging evaluation.

Follow-up PET-CT scans

Five patients had follow-up scans. In two of three NK-cell lymphoma patients in clinical complete remission, PET-CT showed increased uptake in tonsils (SUV = 2.0) and the Fossa of Rosenmuller (SUV = 3.0). These sites were uninvolved at diagnosis, and random biopsies did not show disease infiltration on morphological examination and EBER-ISH. As both cases had received radiotherapy to those sites, chronic infection was a common sequel, which might account for the FDG uptake. Another patient with ALCL had two follow-up scans, both showing good response to treatment. Finally, an AILT patient without palpable disease showed nodal, splenic and marrow FDG avidity, consistent with persistent disease.

Clinical outcome

At a median follow-up of 8 months (range, 1–12 months) since PET-CT, 16 patients were in complete remission. Eight patients (NK-cell lymphoma, n = 4; AILT, n = 2; PTCL-U, n = 2) with advanced stage III/IV disease had died. Five patients were alive with disease (AILT, n = 2; MF, n = 2; T-LGL leukaemia, n = 1). Statistical analysis was performed on 26 patients treated with curative intent. Three patients with chronic refractory disease (MF, n = 2; T-LGL leukaemia, n = 1) and one patient with remitting cutaneous ALCL were excluded. Age, lactate dehydrogenase and advanced stage (III/IV) were significantly correlated with survival and remission status, but the SUVmax on PET had no impact (Table 2). Subset analysis of 12 patients with NK-cell lymphomas also showed that, with plasma EBV DNA, LDH and SUVmax as potential biomarkers for lymphoma load, only EBV DNA and LDH were significantly correlated (r = 0.774; p = 0.003). SUVmax was unrelated to EBV DNA or LDH.

Table 2 Prognostic indicators in 26 cases of mature T-cell and natural killer cell lymphomas
Full size table

Discussion

This study showed a number of interesting findings. Although the role of PET scan in the staging and follow-up of Hodgkin lymphoma and B-cell lymphomas has been well-described [3, 12], data on its clinical utility in mature T-cell and NK-cell neoplasms are limited, with only three series reported [12–14]. In two Western series comprising 22 patients, one series reported only nine patients with predominantly PTCL-U [12], whereas another series reported only MF patients (Table 3) [13]. Data on other subtypes of T-cell and NK-cell lymphomas were poorly represented. A recent Japanese study reported 41 patients, with a skew towards describing patients with PTCL-U [14]. Our study showed a spectrum of T-cell and NK-cell lymphomas typical of the East. There was a predominance of NK-cell lymphomas, comparable numbers of AILT, ALCL and PTCL-U, and fewer cases of MF, a pattern similar to larger number of patients previously reported in these populations [7].

Table 3 PET scans in mature T-cell and natural killer cell neoplasms
Full size table

Our series contained to date the largest number of NK-cell lymphomas investigated by PET-CT. In this lymphoma that presents predominantly in the nasal area, the prognosis is heavily dependent on the initial staging, and hence the International Prognostic Index [15]. Patients with strictly stage I/II nasal NK-cell lymphomas have favourable prognosis with chemotherapy and radiotherapy. However, patients with stage III/IV diseases have a poor outlook. Furthermore, systemic relapses are the main reason for treatment failures in stage I/II patients, indicating that many such patients might have occult distant involvement [16]. Therefore, delineating possible systemic spread is of paramount importance in the initial staging. The current study showed that all nasal NK-cell lymphomas were FDG-avid. Similarly, in extranasal cases, PET-CT defined the extent of dissemination and the presence of occult nasal primary lesions. Similar results were shown by Kako et al. [14] in a smaller number of patients. Moreover, by determining the functional tumour load, PET scan aids in the measurement of the radiation target volume, an important parameter in the planning of radiotherapy [17]. PET may offer significant advantages over CT or magnetic resonance imaging, which often fails to differentiate between soft tissue swellings due to necrosis or inflammation and lymphomatous involvement [18].

AILT and ALCL were also shown to be FDG-avid, which were comparable to results previously reported in fewer patients [12, 14]. Hence, the overall results suggest that PET-CT is reliable in the staging of these mature T-cell lymphomas. All our patients with PTCL-U were FDG-avid, reiterating the findings by Kako et al. [14] in Japanese patients. However, only two of five Caucasian patients with PTCL-U were reported to be PET-positive [12]. As PTCL-U is a heterogeneous group of lymphomas, more cases need to be investigated to define if PET may be more widely applicable in patients of different ethnicities.

The role of PET scan in low-grade T-cell neoplasms is less well-defined. It has been proposed that FDG avidity correlates with aggressive histology [19]. In this study, we also found that indolent T-cell lymphomas showed lower FDG uptake. However, PET might still be useful in defining occult nodal involvement and aggressive transformation in MF [13]. In the only case of T-LGL leukaemia investigated, involved tissues showed no increased FDG uptake, similar to that observed in indolent B-cell lymphoproliferative diseases [20, 21].

The ability of PET in detecting marrow involvement has been evaluated extensively for Hodgkin lymphoma and B-cell lymphomas. In a meta-analysis of 587 cases [21], it was found that the overall sensitivity and specificity of PET for marrow involvement, as compared with bilateral marrow biopsies, were 51% and 91% respectively. There were significant differences in PET detection of marrow involvement in different lymphomas, being best in Hodgkin lymphoma. For B-cell lymphomas, the overall sensitivity and specificity were 43% and 88% respectively. PET identified marrow involvement in 16 of 21 cases (76%) of aggressive B-cell lymphomas, but only 16 of 52 cases (30%) of indolent B-cell lymphomas [21]. In our series, bone marrow biopsy and PET findings were concordant in 26 patients (five of whom had marrow involvement), and discordant in four patients. Of the discordant cases, two patients with NK-cell lymphomas had marrow involvement not shown on PET, but detected by EBER-ISH, a much more sensitive test for marrow involvement as compared with histology [11, 22]. PET also failed to detect marrow involvement in a case of T-LGL leukaemia, an indolent T-cell neoplasm. Conversely, in one case of PTCL-U where trephine biopsies were normal, marrow involvement was detected by PET. A discordance of 14.3% between marrow histology and PET was also shown by Kako et al. [14]. In contrast to our findings, however, their discordant cases were all false negative PETs in aggressive lymphomas such as PTCL-U and AILT. In order to resolve the issues of possible false-positive PET results in small lymph nodes, and false-negative PET in significant lymph nodes, correlative studies of FDG avidity and histological examination will have to be performed.

Finally, most previous studies on the prognostic value of PET adopted qualitative visual assessments (positive or negative scans), with only limited data regarding the value of semi-quantitative assessment with SUV. Hutchings et al. [23] evaluated SUVmax in any region or organ after an interim of two cycles of chemotherapy in patients with Hodgkin lymphoma, and showed that it was an independent prognostic factor negatively affecting the progression-free-survival. However, in our retrospective cohort of patients, interim PET after two courses of treatment had not been performed. Owing to this limitation, it has not been possible for us to evaluate the impact of follow-up PET on treatment outcome. Other studies have addressed the role of initial FDG uptake in predicting chemotherapy response and outcome, but conclusive data have not been reached [24, 25]. In our current series, although prognostic indicators including age, LDH level and advanced stage were significant, SUVmax had no impact on survival or remission. This might be related to the heterogeneity of lymphomas in this series, and the fact that many cases were disseminated, so that SUVmax at one anatomical site might not be representative. However, even in the more localised nasal NK-cell lymphomas, we were unable to show any correlation of SUVmax to LDH and EBV DNA, both valid surrogate markers of tumour load in this lymphoma [10]. It remains to be determined whether the evaluation of total lesion glycolysis, as proposed by Larson et al. [26], might be a more meaningful measure of lymphoma load or response to treatment for mature T-cell and NK-cell lymphomas.