Imaging findings of immune checkpoint inhibitor associated pancreatitis

Abstract

Purpose:

To describe contrast-enhanced computed tomography (CECT), 18-Fluorine (¹⁸F)-fluorodeoxyglucose (FDG) positron emission tomography (PET)/CT and magnetic resonance imaging (MRI) findings of immune checkpoint inhibitor (ICI)associated pancreatitis in patients undergoing immunotherapy for solid malignant tumours.

Method:

In this retrospective study, 25 patients with clinical and/or biochemical evidence of pancreatitis who underwent CECT, MRI and ¹⁸F-FDG-PET/CT while being treated with ICIs were included. Imagingfeatures of acute pancreatitis included: pancreatic enlargement, heterogeneous enhancement, peripancreatic stranding, fluid collection, pseudocyst, necrosis, atrophy and calcification. ¹⁸F-FDG PET/CT imaging was reviewed for pattern of abnormally increased pancreatic FDG uptake. ICI-associated pancreatitis diagnosis was based on clinical, imaging and biochemical findings.

Results:

Imaging findings of ICI-associated pancreatitis included diffuse (n=14) or focal (n=11) pancreatic enlargement; heterogenous enhancement (n=21); focal (n=9) or diffuse (n=15) peripancreatic infiltration on CECT and MRI. A pattern consistent with acute interstitial pancreatitis was present in 20/25 (80%) patients, and a pattern consistent with autoimmune pancreatitis in 4/25 (16%). A mixed pattern was present in one patient (4%). No patient developed necrotizing pancreatitis or a pseudocyst. The CT severity index was < 3 in all patients, consistent with mild pancreatitis. Focal pancreatic FDG uptake was noted in 2/3 (66%) of patients. Acute imaging findings resolved with treatment in all 25 patients. Pancreatic atrophy developed in 11/25 (44%).

Conclusions:

ICI-associated pancreatitis typically presents as either focal or diffuse acute interstitial pancreatitis. Post-pancreatitis atrophy is common. The ICI-associated pancreatitis cases in our study were mild, managed conservatively and did not result in local acute complications.

1. Introduction

Immunotherapy with checkpoint inhibition is a novel oncologic treatment strategy which aims to provoke an antitumor response by impairing inhibitory T-cell pathways, enabling an enhanced immune-mediated response to select malignancies. There have been numerous trials demonstrating the clinical benefit of this treatment strategy, ultimately leading to Food and Drug Administration (FDA) approval for several checkpoint inhibitors in multiple cancer subtypes. [1–5] Examples of checkpoint inhibitors which are in routine clinical use include ipilimumab, which targets cytotoxic T lymphocytic antigen (CTLA-4), pembrolizumab and nivolumab, which target programmed cell death protein 1 (PD-1), and atezolizumab and durvalumab, which target programmed death ligand 1 (PD-L1). [4]

Despite the well described clinical benefits associated with these novel immunotherapy agents, enhancement of normal immunity can result in unwanted immune related adverse events (irAEs), due to the dampening of normal protective immune tolerance. It is important for radiologists to become familiar with the radiological appearance of irAEs as the prompt detection of clinically significant adverse events has the potential to change patient management. [3] The imaging features of multiple organ system irAEs have been described. [6] Those most frequently encountered on imaging include colitis [7, 8] hepatitis [9], pneumonitis [10], and hypophysitis. [11]

A less commonly encountered irAE is ICI-associated pancreatic injury which can result in acute pancreatitis. [3, 6, 12] ICI-associated pancreatitis has a reported incidence of 1–2% [5, 12] and may rarely lead to both endocrine and exocrine dysfunction. [13] Limited data exist describing the imaging features of ICI-associated pancreatitis with a few reports describing imaging manifestations of both focal and diffuse pancreatic involvement. [12, 14–17] The objective of this article is to discuss the imaging findings of ICI-associated pancreatitis and identify any potential long-term imaging and clinical sequalae, based on a systematic review of longitudinal cross-sectional imaging during therapy.

2. Materials and Methods

2.1. Patient population and clinical assessment

Institutional review board approval was obtained and the requirement for informed consent was waived for this retrospective review. Studied patients were treated with ICIs between August 15, 2016 and January 9, 2019 and were identified from a prospectively maintained educational databaseof radiological irAEs. Clinical and biochemical parameters were obtained from the hospital radiology information system (RIS). The date of onset of clinical symptomatology (new onset of abdominal pain, nausea, vomiting) and autoimmune disease history were recorded, if present. Risk factors for acute pancreatitis were also noted including cholelithiasis, excessive alcohol consumption (based on physician clinical evaluation), corticosteroid history and any prior history of abdominal trauma or biliary intervention.

Inclusion criteria were based on guidelines from the American College of Gastroenterology for the diagnosis of acute pancreatitis [18]: (i) imaging evidence of pancreatitis, in addition to at least one of the following features: (a) abdominal pain (b) elevated serum lipase or amylase three or more times the upper limit of normal [18] (ii) ICI treatment at the time of presentation with pancreatitis (iii) absence of another clinically demonstrable etiology for pancreatitis. Twenty five patients who met these criteria were included in the study. Baseline patient demographics are described in Table 1.

Table 1.

Baseline demographics of patients with immune checkpoint inhibitor-associated pancreatitis

Patients with ICI-associated pancreatitis (n=25)
Age (years)		57 (range 45–85)
Gender
	Female	14
	Male	11
Primary cancer
	Melanoma	16
	Endometrial	2
	Lymphoma	2
	Lung	2
	Renal	2
	Ovarian	1
History of prior or concomitant autoimmune disease		6
	Hypothyroidism	4
	Raynauds	1
	Sjogrens	1
History of prior or concomitant immune related adverse event		11
	Hypophysitis	4
	Appendicitis	2
	Nephritis	1
	Colitis	1
	Thyroiditis	1
	Optic neuritis	1
	Pneumonitis	1

2.2. Cross-sectional imaging acquisition

Multidetector contrast enhanced computed tomography (CECT) images of the abdomen (Lightspeed 16 and VCT, GE Medical Systems) were obtained following intravenous (IV) contrast (Omnipaque 300, 300 mg/I/mL) with a volume of 150 mLs, administered at a rate of 2–3mLs. Helical imaging was performed during portal venous phase at 60–80 seconds at 120 kVp, minimum mAs of 100–150 with 5mm slice thickness. Preceding 18-Fluorine (¹⁸F)-fluorodeoxyglucose (FDG) positron emission tomography (PET)/CT, all patients were kept nil per os for at least 4 hours and serum glucose was measured to ensure a normal range of values. Patients were then injected with 0.15–0.2 mCi/kg of ¹⁸F-FDG and PET images were acquired from the vertex of the skull to the mid thighs, without IV contrast. Magnetic resonance imaging (MRI) was performed using the following protocol: T2-weighted imaging (WI) single shot fast spin echo (SSFSE) in both axial and coronal planes, T1WI in-phase and out-phase gradient echo (GE) in axial plane with 5 mm slice thickness and T2WI magnetic resonance cholangiopancreatography (MRCP) with 3 mm or less slice thickness and 3D volumetric acquisition in coronal plane. T1 dynamic fat-saturated contrast-enhanced MRI was obtained following IV gadobutrol (Gadovist, Schering, Berlin, Germany) 0.1 mmol/kg of the patient’s body weight at 1.5 ml/s injection rate followed by a 20ml normal saline bolus flush. Dynamic scanning was performed including arterial (20–30 seconds), pancreatic (35–50 seconds) and portal venous (60–70 seconds) imaging following administration of contrast agent.

2.3. Imaging assessment

Retrospective review of CECT, MRI abdomen examinations and ¹⁸F-FDG PET/CT exams was performed on a dedicated picture archiving and communication system (PACS) workstation. The Atlanta and HISORt criteria [19, 20]were used to evaluate CECT images for the features of acute pancreatitis and autoimmune pancreatitis respectively. The imaging features assessed as per the Atlantacriteria were: presence of (i) pancreatic enlargement (focal or diffuse) (ii) enhancing but heterogenous parenchyma (focal or diffuse) (iii) peripancreatic stranding, (iv) acute peripancreatic fluid collection, (v) acute necrotic collections (vi) pancreatic necrosis, (vii) peripancreatic necrosis, (viii) main pancreatic duct dilatation. The imaging features assessed as per the HISORt criteria were: (i) diffusely enlarged gland with rim enhancement, (ii) diffusely irregular/attenuated main pancreatic duct, (iii) focal pancreatic enlargement, (iv) focal pancreatic duct stricture, (v) pancreatic atrophy, (vi) pancreatic calcification.After the individual features of pancreatitis were documented, each patient was classified as having an imaging pattern consistent with acute interstitial pancreatitis or autoimmune pancreatitis. The pattern and distribution of abnormal pancreatic FDG uptake was also noted. In addition, the presence of pancreatic atrophy was recorded, where present, following resolution of acute pancreatitis and was classified as a visually estimated size reduction (mild <25%; moderate 25–50%; severe >50%). For each patient, the same cross-sectional imaging modality was used for the baseline pancreatic size and atrophy assessment.

Independent imaging assessment was performed by a body imaging radiologist with 5 years of experience, and a body imaging fellow. Where assessment was discordant, a third body radiologist with > 10 years’ experience (author 4) reviewed the imaging and the case was discussed until a consensus was achieved. No predetermined schedule for imaging follow-up was adhered to, with resolution of pancreatitis and development of atrophy noted on surveillance imaging performed for the evaluation of metastatic cancer. The presence of pancreatic or peripancreatic metastases was also recorded.

3. Results

At the time of symptomatology or biochemical evidence of pancreatitis, all 25 patients underwent CECT. In addition, three patients were imaged with ¹⁸F-FDG PET/CT and a single patient was imaged using MRI.

3.1. Imaging findings

The mean interval from onset of symptoms or first noted elevation of serum pancreatic enzymes above threshold levels to imaging was 5.7 days (range 1– 36 days). Of the 17/25 (68%) with clinical symptoms of pancreatitis, the median time from onset of symptomatology to index imaging was 2.3 days (range 1–7) while those with biochemical suspicion for pancreatitis (32%) were imaged at a median of 8.4 days (range 1–36) from initial abnormal biochemical laboratory result.

The clinical and imaging features of all patients are summarized in Table 2. The most common CECT finding of ICI-associated pancreatitis was pancreatic enlargement, seen in all patients. In 14/25 (56%) diffuse pancreatic enlargement was exhibited (Figure 1) with 11/25 (44%) demonstrating focal pancreatic enlargement (Figure 2). Mass-like enlargement was seen in four patients, including the patient imaged with MRI (Figure 3). In all cases of mass-like pancreatic enlargement, resolution of focal swelling was observed on imaging follow-up. No patient underwent tissue sampling to evaluate for metastatic disease or the interval development of a primary pancreatic malignancy. Heterogenous enhancement was identified in 21/25 (84%) of which 12/21 (57%) demonstrated diffusely heterogenous enhancing parenchyma. Peripancreatic fat stranding was noted in 21/25 (84%) of patients, 10/21 (48%) developed diffuse fat stranding and 11/21 (52%) developed focal fat stranding (Figure 4). Focal dilatation of the main pancreatic duct was seen in a single patient. 11 patients (44%) developed post-pancreatitis parenchymal atrophy; five patients had mild atrophy, with two developing moderate atrophy and four severe atrophy (Figure 5).No patients developed a peripancreatic fluid collection, pancreatic necrosis, portal or splenic vein thrombosis, gastric outlet obstruction or colonic necrosis.

Table 2.

Clinical andimagingcharacteristics of cases of immune checkpoint inhibitor associated pancreatitis

Patient no.	Immune checkpoint inhibitor (ICI)	Prior Immunotherapy	Abdominal pain	Elevated enzymes	Imaging findings		Time from onset to imaging (days)	ICI d/c***	Treatment	Time to resolution on imaging (days)
					Pattern	Severity*
1	Pembrolizumab	No	No	Yes	Interstitial oedematous	1	13	Yes	IV/oral steroids	168
2	Nivolumab	Yes	Yes	Yes	Interstitial oedematous	2	0	Yes	Oral steroids	89
3	Nivolumab	Yes	Yes	Yes	Interstitial oedematous	2	10	Yes	Oral steroids	90
4	Pembrolizumab	No	Yes	Yes	Mixed pattern**	2	1	Yes	IV/oral steroids, Infliximab	51
5	Atezolizumab	No	No	Yes	Interstitial oedematous	2	0	Yes	Oral steroids	94
6	Nivolumab	No	No	Yes	Interstitial oedematous	2	0	Yes	Oral steroids	58
7	Nivolumab	Yes	Yes	Yes	Interstitial oedematous	1	2	Yes	Oral steroids	177
8	Pembrolizumab	No	Yes	Yes	Autoimmune	2	2	Yes	Oral steroids	129
9	Ipi/Nivo****	No	Yes	Yes	Autoimmune	2	6	Yes	IV/oral steroids, MM+	164
10	Nivolumab	No	No	Yes	Interstitial oedematous	1	36	Yes	Oral steroids	152
11	Ipi/Nivo	No	No	Yes	Interstitial oedematous	2	17	Yes	Oral steroids	40
12	Nivolumab	Yes	Yes	Yes	Interstitial oedematous	2	2	Yes	Oral steroids	24
13	Pembrolizumab	Yes	Yes	No	Interstitial oedematous	2	1	Yes	Oral steroids	122
14	Ipi/Nivo	Yes	Yes	Yes	Interstitial oedematous	2	1	Yes	Oral steroids	68
15	Pembrolizumab	No	No	Yes	Interstitial oedematous	2	0	Yes	Oral steroids	64
16	Ipi/Nivo	Yes	Yes	No	Autoimmune	2	1	Yes	Oral steroids	83
17	Ipi/Nivo	No	Yes	Yes	Autoimmune	2	0	Yes	Oral steroids	88
18	Pembrolizumab	Yes	Yes	Yes	Interstitial oedematous	2	2	Yes	Oral steroids	182
19	Nivolumab	Yes	Yes	No	Interstitial oedematous	2	0	Yes	Oral steroids	24
20	Pembrolizumab	No	Yes	Yes	Interstitial oedematous	2	1	Yes	Oral steroids	78
21	Ipi/Nivo	No	Yes	Yes	Interstitial oedematous	2	2	Yes	Oral steroids	76
22	Pembrolizumab	Yes	No	Yes	Interstitial oedematous	2	0	Yes	Oral steroids	201
23	Pembrolizumab	Yes	No	Yes	Interstitial oedematous	1	14	Yes	Oral steroids	194
24	Ipi/Nivo	No	Yes	Yes	Interstitial oedematous	2	16	Yes	IV/oral steroids, Infliximab	46
25	Nivolumab	Yes	No	Yes	Interstitial oedematous	2	12	Yes	Oral steroids	74

^*As per Balthazar Computed Tomography severity index

^**Interstitial oedematous with focal pancreatic duct dilatation suggestive of autoimmune

^***Discontinued

^****Ipilimumab/ Nivolumab

⁺Mycophenolate mofetil

Figure 1.

47-year-old male with melanoma before (A) and following (B) treatment with nivolumab.

A. Coronal CT image of the upper abdomen after intravenous contrast demonstrates a normal pancreatic body and head (arrowheads).

B. Coronal CT image of the upper abdomen after intravenous contrast in the same patient after the initiation of checkpoint inhibitor immunotherapy demonstrates new diffuse enlargement of the pancreatic body and head (arrowheads) and new diffuse peripancreatic fat stranding (arrows). This pattern was classified as acute interstitial pancreatitis.

Figure 2.

70-year-old-female with melanoma treated with nivolumab. Axial CT image of the upper abdomen after intravenous contrast after the initiation of checkpoint inhibitor immunotherapy demonstrates focal pancreatic tail enlargement and subtle peripancreatic fat stranding. This pattern was classified as acute interstitial pancreatitis.

Figure 3.

50-year-old-female with metastatic lung adenocarcinoma following treatment with pembrolizumab.

A. Axial MR image of the upper abdomen following intravenous gadolinium after initiation of immune checkpoint inhibitor therapy shows mass-like enlargement of the pancreatic head.

B. Coronal MR image of the upper abdomen following intravenous gadolinium at the same timepoint as the image shown in Figure 3A demonstrating abrupt cut-off of the common bile duct. This pattern was classified as autoimmune pancreatitis.

Figure 4.

64-year-old-female with lung adenocarcinoma before (A) and following (B) treatment with nivolumab.

A. Coronal CT image of the upper abdomen after intravenous contrast demonstrates a normal pancreatic head/uncinate process (arrowheads) and normal fat interposed between the pancreas and duodenum (arrows).

B. Coronal CT image of the upper abdomen after intravenous contrast after the initiation of checkpoint inhibitor immunotherapy demonstrates subtle mild enlargement and heterogenous enhancement of the pancreatic parenchyma (arrowheads), and very subtle stranding of the fat interposed between the pancreas and duodenum (arrows). In cases with subtle findings such as this, knowledge of the clinical suspicion for ICI associated pancreatis is vital. This pattern was classified as acute interstitial pancreatitis.

Figure 5.

- 70-year-old-female with melanoma treated with nivolumab. Axial CT image of the upper abdomen after intravenous contrast in the same patient as Figure 2 (obtained 7 months after the timepoint depicted in Figure 2) demonstrates marked diffuse pancreatic atrophy (arrow).

Of the 3 patients who underwent additional ¹⁸F-FDG PET/CT imaging, two (66%) demonstrated focal FDG uptake in the body and tail with maximum standardized uptake values (SUV_max) of 3.4 and 3.6 respectively. Generalised increased FDG uptake throughout the entire pancreas was seen in one patient (33%) (Figure 6).

Figure 6.

56-year-old female with ovarian cancer following treatment with ipilimumab and nivolumab. Axial fused 18F-FDG PET/CT image of the upper abdomen after the initiation of checkpoint inhibitor immunotherapy demonstrates diffuse intense uptake in the pancreas (arrow).

Based on the assessments above, patients were classified into two distinct patterns of ICI-associated pancreatitis. The majority (80%) demonstrated a pattern of acute interstitial pancreatitis, primarily characterised by non-mass like pancreatic swelling, heterogeneous enhancement and peripancreatic inflammation. Four (16%) patients exhibited features of autoimmune pancreatitis with focal mass-like pancreatic enlargement. A single patient showed a mixed phenotype with predominantly interstitial pancreatitis in addition to focal main pancreatic duct dilatation, more commonly seen in autoimmune pancreatitis. No other features associated with autoimmune pancreatitis, as described in the HISORT criteria [20], were identified in our cohort.

At time of initial imaging of ICI-associated pancreatitis, the majority of patients (21/25 [68%]) demonstrated a score of 2/10 on the Balthazar CT severity index (CTSI). [18, 21] No patient met the criteria for a CTSI score of 3 or more, and thus all were classified as having mild pancreatitis. No patient developed necrotizing pancreatitis.

3.2. Clinical features

22/25 (88%) patients met the biochemical criteria for acute pancreatitis with elevated serum lipase and/or amylase, while 17/25 patients (64%) experienced abdominal symptomatology consistent with acute pancreatitis. A prior history of irAE was seen in 11/25 (44%) patients; the most common prior irAE was hypophysitis in four (16%) patients. 6/25 patients (24%) had a pre-existing autoimmune condition unrelated to ICI therapy, including hypothyroidism (n=4), Sjogren’s syndrome (n=1) and Raynaud’s disease (n=1).

Eighteen (72%) patients were receiving ICI monotherapy at the time of diagnosis of acute pancreatitis, with almost all (17/18) being treated with anti-PD1 therapy, 9/18 (52.9%) receiving nivolumab, with the remaining 8 patients receiving pembrolizumab. Of these patients, 11/18 had radiologic evidence of focal pancreatitis, the majority (8/11) in the tail. A single patient was being treated with atezolizumab alone. Combination therapy with nivolumab and the anti-CTLA agent ipilimumab was used for the remaining seven patients, with four developing diffuse pancreatitis and three focal tail pancreatitis. The mean duration of ICI treatment in our patient cohort was 96 days (range 33–232 days).

3.3. Risk factors for pancreatitis

Of the 25 patients, two (8%) had cholelithiasis, however none had evidence of acute biliary pathology. Three patients had a prior cholecystectomy. 12 (48%) described occasional consumption of alcohol with all remaining patients reporting no history of alcohol intake. Five patients were receiving a low dose oral corticosteroids at time of suspected ICI-associated pancreatitis because of a prior irAE. No patients had a history of biliary instrumentation. IgG4 levels were checked in 4/25 patients and were within normal limits (n=3) or below normal (n=1). Of note, mass-like enlargement was not noted in any of these four patients.

3.4. Management and Outcomes

The mean time interval from initiation of ICI therapy to the development of acute pancreatitis was 89 days (range: 20–330 days). Twenty-one (84%) were managed with oral corticosteroid therapy and discontinuation of immune checkpoint inhibitor. Four patients (16%) with concomitant optic neuritis, colitis and hyophysitis (n=2) respectively received IV methylprednisolone and IV fluids in addition to infliximab or mycophenolate.

Resolution of the acute imaging signs of ICI-associated pancreatitis occurred on follow-up CT imaging in all 25 patients. Resolution was seen in 11/25 patients on first initial follow-up CT scan, and on the second or subsequent follow-up imaging study in the remaining 14 patients. Mean time from onset of acute pancreatitis to resolution on imaging was 101 days (range 24–201 days). ICI treatment was discontinued permanently in 11/25 patients. No patients required percutaneous image-guided drainage or open surgical intervention.

11 patients developed pancreatic atrophy following resolution of ICI-associated pancreatitis. 4/11 (36%) of these patients developed pancreatic dysfunction. Three (27%) developed exocrine dysfunction manifesting as diarrhoea or steatorrhea initially attributed to suspected ICI-associated colitis, that failed to resolve with steroids. Of these patients, two demonstrated severe atrophic change. Patients with pancreatic atrophy were imaged for a median of 14 months (range 7–33 months). The median time interval from onset of pancreatitis (clinical or biochemical) to exocrine dysfunction was 244 days (range 103–311 days). One (9%) patient with moderate pancreatic atrophy developed new onset endocrine dysfunction (diabetes mellitus [DM]). One additional patient developed DM but did not demonstrate atrophy on imaging. Endocrine dysfunction presented between three and five months following onset of clinical pancreatitis, and in both cases required treatment with oral hypoglycaemics and subcutaneous insulin. Of the five total patients who developed pancreatic dysfunction following ICI-associated pancreatitis, three presented with focal pancreatitis while two patients demonstrated diffuse pancreatitis. Three of five received combination therapy with ipilimumab and nivolumab while two received monotherapy (treated with nivolumab and pembrolizumab respectively). The mean duration of ICI treatment was 114 days in the 5 patients with post-pancreatitis pancreatic dysfunction. The Balthazar CT severity index was 2 in all 5 of these patients.

4. Discussion

Checkpoint inhibition and immunomodulation represents an important new frontier in the field of oncology, however the inflammatory adverse events associated with ICI therapy pose a challenge for clinical management. Pancreatitis is one such rare complication of ICI treatment for solid malignant tumours. [17, 22] We have described the CT imaging features of ICI-associated pancreatitis in a cohort of patients receiving immunotherapy for a variety of solid malignancies. An imaging pattern of mild acute interstitial pancreatitis was observed in the majority of cases (80%). No patient developed acute complications, however 44% of patients developed chronic pancreatic atrophy, of whom 36% developed pancreatic dysfunction. Resolution of acute radiological findings following discontinuation of the ICI agent and initiation of systemic steroid therapy was noted at a mean of 101 days on follow-up imaging.

ICI side effects are common, affecting up to 60% of patients, and can be graded according to clinical severity as mild (grade 1), moderate (grade 2), severe (grade 3) or life-threatening (grade 4) based on the Common Terminology Criteria for Adverse Events (CTCAE). [23–26]A recent systematic review and meta-analysis found that 0.9% of patients died from irAEs, usually due to bowel perforation as a complication of ICI-associated colitis. [27]The most common radiologically encountered irAEs include colitis, pneumonitis, hypophysitis and sarcoid-like reaction with up to 31% of patients noted to have adverse events on body imaging studies over the course of their ICI treatment. [3, 5, 28–30] Overall incidence of grade 2 ICI-associated pancreatitis is less than 5%. [12, 31, 32] Chronic pancreatic dysfunction may occur but no deaths attributable to ICI-associated pancreatitis have been reported to date. [12, 33–35]

In the few previous reports of ICI-associated pancreatitis, both focal and diffuse patterns of pancreatic inflammation have been described on CT. [2, 5] An autoimmune pancreatitis pattern has also been reported, in a patient with “sausage shaped” pancreatic enlargement and no peripancreatic stranding.[6] Abuh-Sbeih et al. noted that the most common CT imaging features of ICI-associated pancreatitis were peripancreatic fat stranding, pancreatic enlargement with heterogenous enhancement, and segmental hypoenhancement. Three of their cohort developed pancreatic atrophy on follow up imaging. 15% percent of patients developed long-term adverse outcomes related to ICI-associated pancreatitis, including recurrent pancreatitis and DM. [13] On PET/CT, both focal and diffuse patterns of increased FDG uptake have been reported [16, 35] Cases with focal uptake can indeed mimic metastatic disease.[16]

To the best of our knowledge, this is the largest patient cohort with systematic description of the imaging of ICI-associated pancreatitis. In our cohort, two recognizably different radiological patterns of pancreatitis were observed. The first pattern, seen in the majority (80%) of cases, was characterized by focal or diffuse pancreatic enlargement, peripancreatic fat-stranding and heterogenous enhancement, most characteristic of acute interstitial pancreatitis. A second pattern of mass-like enlargement (often seen in autoimmune pancreatitis) was present in a minority (16%) of patients. One patient developed a mixed pattern of disease on CT, with both interstitial oedematous and autoimmune features.

The imaging assessed severity of pancreatitis in our cohort was mild, with no patient scoring ≥3 on the CT severity index. We noted that the majority of those patients receiving ICI monotherapy (11/18) developed a focal pancreatitis, whereas most of those receiving combination therapy (4/7) presented with diffuse pancreatitis, in keeping with some prior reports suggesting that more severe irAEs can develop when a combination of drugs is given (George). 44% developed pancreatic atrophy, including 16% who developed severe atrophy. Five (20%) of our patient cohort developed exocrine or endocrine dysfunction with four demonstrating pancreatic parenchymal atrophy on follow-up CT imaging, compared to baseline studies. 64% of patients with pancreatic atrophy did not develop exocrine or endocrine dysfunction. For the five total patients who developed pancreatic dysfunction, two demonstrated a diffuse pancreatitis on imaging and presentation and received ICI treatment for a measn duration of 114 days. Interestingly, Abu Sbeih et al. observed that patients who developed adverse outcomes of ICI-asociated pancreatitis received a longer duration ICI therapy (mean 412 days) than patients who did not (mean 200 days). (Abu Sbeih)

Pancreatic atrophy is well described following alcohol-induced pancreatitis and autoimmune pancreatitis and may result in synthetic dysfunction. [36–38] The association between pancreatic atrophy and pancreatic insufficiency has also been explored in ICI-treated patients. Tan et al. conducted a scoping review of ICI-induced endocrinopathies and identified 66 cases of ICI-induced diabetes mellitus of which three had pancreatic atrophy and two had pancreatitis. [31] Eshet et al. noted a 7.7% (31/403) incidence of pancreatic atrophy in patients receiving ICI therapy. Of these patients, 13% (4/31) developed exocrine pancreatic insufficiency. The presence of atrophy radiologically did not correlate with clinical symptoms of exocrine dysfunction but did correlate with a significantly reduced overall survival. [34]

Knowledge of the imaging findings of ICI associated pancreatitis has several important clinical implications. Cross-sectional imaging is an important first step in the initial work up of patients presenting with clinical and/or biochemical evidence of pancreatitis while receiving ICI treatment. In our cohort, the features of ICI-associated pancreatitis were often subtle on imaging and that comparison to baseline pancreas morphology to detect mild pancreatic enlargement and peripancreatic inflammatory change is useful in patients when clinical suspicion is high. This is particularly true in patients with minimal or no symptoms, but who have biochemical evidence of pancreatic injury. In our analysed group, approximately one third of patients were asymptomatic at the time of imaging, presenting only with elevated enzymes and without reported pain. In a retrospective review of 119 patients with melanoma receiving nivolumab and ipilimumab, Friedman et al. observed that only 20% of patients with elevated serum pancreatic enzymes (grade 3 or higher) had pancreatitis and concluded that in patients treated with ICIs, an elevation in amylase and lipase could reflect inflammation of organs other than the pancreas that produce these enzymes. [24] This phenomenon emphasizes the importance and potential role of imaging in identifying asymptomatic patients with raised serum pancreatic enzymes who may not be suspected to have ICI-associated pancreatitis based on clinical assessment alone. The natural history of untreated ICI-associated pancreatitis remains unclear, but from prior reports asymptomatic patients appear to have a relatively higher rate of long term sequalae such as chronic malabsorption and DM following ICI-associated pancreatic injury, possibly due to undertreatment of the acute pancreatic insult. [13]

In this context, reviewing cases of suspected immune related toxicity in multidisciplinary team meetings may be beneficial, as this environment allows clinical and radiological finding to be synthesised. The early identification of acute pancreatitis allows for prompt treatment with steroids, and close monitoring for clinical and radiological improvement. A prompt diagnosis of ICI-associated pancreatitis also informs the decision of whether to discontinue immunotherapy. Finally, we suggest that reviewing CTs for evidence of pancreatic atrophy may be a useful method for evaluating for prior episodes of clinically occult pancreatic injury in patients with exocrine or endocrine pancreatic dysfunction following ICI therapy. [12]

There are several limitations to our study. Firstly, the design was retrospective and does not allow us to estimate a true incidence of ICI-associated pancreatitis. Second, some patients had other risk factors for pancreatitis, but in every case included in this study effort was made to exclude these other potential aetiologies as the cause for the episode of pancreatitis. Thirdly, the size of the cohort in this study is relatively small, but in light of the rarity of this complication we believe the reporting of 25 cases is an important addition to the literature as the largest case series available. Prospective studies with larger sample sizes are needed to more thoroughly evaluate the association between radiological features and clinical outcomes in ICI-associated pancreatitis and whether other imaging modalities of the pancreas will further help define this condition.

In conclusion, two distinct radiologic patterns of ICI-associated pancreatitis were observed in our patient cohort: a majority presented with findings of acute interstitial pancreatitis. A large number of patients developed pancreatic atrophy several of whom developed exocrine and endocrine dysfunction. As the use of ICI therapy increases, radiologists will more frequently encounter irAEs. Early recognition of ICI-associated pancreatitis on imaging can facilitate timely treatment potentially limiting pancreatic injury.

Highlights:

• This study describes the imaging features of immune checkpoint inhibitor (ICI)-associated pancreatitis

• Interstitial oedematous pancreatitis is the most common pattern of disease

• Pancreatic atrophy following resolution of ICI-associated pancreatitis is common

Abbreviations

CECT

Contrast-enhanced computed tomography

CTCAE

Common Terminology Criteria for Adverse Events

CTSI

Computed tomography severity index

CTLA-4

Cytotoxic T lymphocytic antigen

DM

Diabetes mellitus

¹⁸F-FDG PET/CT

18-Fluorine fluorodeoxyglucose positron emission tomography

FDA

Food and Drug Administration

GE

Gradient echo

ICI

Immune checkpoint inhibitor

irAEs

Immune related adverse events

IV

Intravenous

MRCP

Magnetic resonance cholangiopancreatography

MRI

Magnetic resonance imaging

PACS

Picture archiving and communication system

PD-1

Programmed cell death protein 1

PD-L1

Programmed death ligand 1

RIS

Radiology information system

T1WI

T1-weighted imaging

T2WI

T2-weighted imaging

SSFSE

Single shot fast spin echo

SUV_max

Maximum standardized uptake value