CT imaging of primary pancreatic lymphoma: experience from three referral centres for pancreatic diseases

Primary pancreatic lymphoma (PPL) is a very rare disease, representing only 0.1% of malignant lymphomas and 0.2% of primary pancreatic tumours [1, 2]. The World Health Organisation provides precise diagnostic criteria: the bulk of the disease is localised in the pancreas; adjacent lymph node involvement and distant spread may exist but the primary clinical presentation has to involve the pancreatic gland [3]. The disease can develop at any age, but usually affects elderly patients, with higher incidence in males. Clinical presentation is variable: in most cases the first manifestation is abdominal pain; other clinical findings include systemic symptoms such as fever, night sweats and weight loss. Jaundice and/or gastric or duodenal obstruction may be present [2]. These heterogeneous symptoms may overlap those of other pancreatic tumours, resulting in diagnostic difficulties [4, 5]. Histopathological examinations reveal a framework compatible with non-Hodgkin lymphoma, usually diffuse large B cell lymphoma (DLBCL). Nevertheless, other histological types have been reported including indolent diseases such as follicular and marginal zone lymphoma [3‐6]. Diagnosis can be obtained through a percutaneous/endoscopic US-guided biopsy or exploratory laparotomy biopsy. The correct diagnosis is essential because, unlike epithelial tumours, patients with lymphoma do not benefit from surgery but from chemotherapy and indeed surgery may increase morbidity [5].

Up to now, reports of the CT imaging appearance of PPL are limited to a few papers including mainly case reports [7‐10]. The aim of this study is to describe the imaging characteristics of PPL at CT based on a relatively large series of cases examined in three institutions in three different countries.

Histopathological results were: follicular non-Hodgkin lymphoma in five patients, diffuse large B-cell lymphoma (DLBCL) in six patients and high-grade B-cell lymphoma not otherwise specified in three patients. The replicative Ki67 index score was below 25% in two cases, between 25% and 50% in four cases, between 50% and 75% in one patient and more than 75% in seven cases. The histopathological features are summarised in Table 2.

Qualitative analysis. Six out of 14 PPLs were located in the pancreatic head and 7 in the body-tail and 1 involved the whole gland (Figs. 1 and 2). In 5/14 cases the caeliac axis, superior mesenteric artery and superior mesenteric vein were encased; splenic vein and artery encasement was depicted in two PPLs. Even when the vessels were encased, there were no signs of infiltration, with no vessel wall irregularity or stenoses.

Two patients had tumoral necrosis, involving respectively about 20% and 50% of the pancreatic mass lesion. Enlarged retroperitoneal or mesenteric lymph nodes were depicted in 11/14 cases. Fat stranding in the peri-pancreatic region was observed in all patients. The common bile duct was dilated in 6/14 cases; the main pancreatic duct upstream of the neoplasm was enlarged in 5/14 patients. Qualitative parameters are summarised in Table 3.

Quantitative analysis. Mean neoplasm longest dimension and volume were 8.05 cm and 210.8 cm³. Mean tumour attenuation values were 39.1 HU in the unenhanced scan, 60.6 HU in the pancreatic phase and 71.4 HU in the venous phase (Fig. 3). Quantitative parameters are summarised in Table 4.

Although several case reports of PPLs have been published, there is still little information regarding the imaging features of these rare lesions [4, 5, 7, 8]. The current study encompasses 7 years of experience at three different institutions in three different countries and describes CT and histopathological features of 14 patients with PPL. The rarity of PPL makes it difficult to recognise this neoplasm since it presents characteristics in common with ductal adenocarcinoma, the most frequent pancreatic solid tumour or, more rarely, with hypovascular neuroendocrine tumours. PPL could also be confused with the focal or diffuse forms of autoimmune pancreatitis; all these conditions present different management approaches, with distinct treatment and prognosis [1, 7, 9].

Every patient in our series presented with abdominal pain or discomfort; other frequent clinical manifestations were jaundice, a palpable abdominal mass and systemic symptoms, such as fever, night sweats and weight loss. Unfortunately, pancreatic adenocarcinoma and autoimmune pancreatitis often have the same clinical picture [14]. As confirmed by the current study, the presence of an increased LDH serum level without an increased CA19–9 level favours a lymphomatous aetiology of the pancreatic mass [5]. The final histopathological diagnosis was available for every patient: the majority were histologically aggressive DLBCL or high-grade B-cell lymphoma not otherwise specified, with a high Ki67 index (> 75%). Only 5/14 presented low-grade malignancy (follicular lymphomas with Ki67 < 50%). According to the literature, the most common PPL histological subtype is DLBCL followed by follicular lymphoma. Before the 2016 WHO classification, most high-grade B-cell lymphomas not otherwise specified were included in the DLBCL group. Therefore, our data are in agreement with previous studies [5, 15].

Some particular imaging features in patients with PPL were observed. There was not a predilection for a pancreatic region and the whole gland could be globally involved. Vascular encasement was typical, but significant irregularities in the vessels wall were never found (Fig. 3). This peculiarity could be helpful in the differential diagnosis with ductal adenocarcinoma, which causes vascular infiltration at early stages, with erosion of vessel walls, stenoses and eventually tumoral thrombosis [14]. The pancreatic parenchyma affected by lymphoma has low attenuation values compared to the healthy pancreas; as opposed to adenocarcinoma, areas of necrosis are rare. After contrast medium administration the attenuation is higher in the portal-venous phase compared to the arterial-pancreatic phase (Fig. 4). Enlarged retroperitoneal lymph nodes are frequent and fat stranding is always present. The double duct sign, simultaneous dilation of the common bile and pancreatic ducts, typical in patients with adenocarcinoma of the pancreatic head, was only depicted in 5/14 PPLs [1, 14‐17].

In the most difficult cases other imaging modalities are recommended in addition to CT: MRI, in particular MR cholangiography sequences, can give important information about ductal system morphology (a single stricture of the pancreatic duct with an upstream enlarged duct is highly suggestive of ductal adenocarcinoma) [16]. Evaluation of stenoses in the ductal system could be improved by the administration of secretin to distinguish between the infiltrated main duct (negative duct-penetrating sign after drug administration, typical of adenocarcinoma) and compressed duct (positive duct-penetrating sign, suggestive of a lymphomatous process) [16]. Diffusion-weighted imaging helps to recognise the areas of pancreatic parenchyma involved by the neoplasm and could have a role in the evaluation of treatment response [17]. Nuclear medicine, in particular ¹⁸FDG PET-CT, is indicated to depict involved lymph nodes in the whole body and to assess the metabolic activity of the primary neoplasm. The recently proposed PERCIST system (PET Response Criteria in Solid Tumours) evaluates changes in the standardised uptake value (SUV) of ¹⁸FDG to assess treatment response [18].

Most of the above-mentioned imaging features could allow distinguishing PPL from pancreatic adenocarcinoma, but the differential diagnosis with autoimmune pancreatitis remains difficult. As reported by previous authors, the size of the parenchyma involved by the autoimmune process is smaller than for PPL [3]. In addition, symptoms of autoimmune pancreatitis are less severe and the disease is often associated with increased IgG4 serum levels [19]. According to our experience, it is mandatory to integrate imaging findings with the clinical and laboratory results to make the correct diagnostic hypothesis; histopathological confirmation is always necessary, also to evaluate the lymphoma subtype.

The present study has some limitations, namely, its retrospective design and the relatively small sample size. A larger prospective study would be required to corroborate our findings.

In conclusion, the most suggestive CT signs of primary pancreatic lymphoma are a large hypodense mass lesion with progressive contrast enhancement, absence of necrosis, lack of dilation of the main pancreatic duct and presence of vessel encasement without infiltration.