Diarrhea is one of the most common irAEs, affecting approximately 44% (vs. 10% for grade 3-4) of patients treated with a combination of CTLA-4 and PD-(L)1 inhibitors, 36% (vs. 8% for grade 3-4) of patients treated with anti-CTLA-4, and 11% (vs. 1% for grade 3-4) of patients treated with anti-PD-(L)1 (34). This symptom is often associated with colitis, another common irAE, as it is reported in 16% (vs. 11% for grade 3-4) (combination of CTLA-4 and PD-(L)1 inhibitors), 8% (vs. 5% for grade 3-4) (anti-CTLA-4), and 1% (vs. 1% for grade 3-4) (anti-PD-(L)1) of ICI-treated patients (34). With a median time to full onset of 5 to 10 weeks, colitis can lead to various complications, including intestinal perforation, ischemia, necrosis, hemorrhage, and toxic megacolon (35, 36). Therefore, the typical diagnostic workup in these cases includes contrast-enhanced CT, in which ICI-induced colitis appears as diffuse inflammation with bowel wall thickening (>4mm), mucosal hyperenhancement, mesenteric hyperemia, mesenteric vessel congestion, and air-fluid levels (8, 37). In addition, cases of segmental colitis in association with diverticulosis and isolated rectosigmoid colitis without diverticulosis have been described (38, 39). A recent study including patients with various types of cancer showed CT findings suggestive of colitis in 20 of 34 patients with symptoms of colitis and in 5 of 19 patients even without clinical presentation of colitis (40). ¹⁸F-FDG PET/CT may reveal increased partial or diffuse tracer uptake in colitis or throughout the entire bowel in patients with extensive inflammation (41, 42) ( Figure 1 ). Moreover, it has been reported that it might be more sensitive than CT for the early detection of colitis in patients undergoing ICI treatment (43). However, its lack of specificity for instance due to physiological muscular activity or in patients treated with metformin hinders its value in routine practice (41, 43).

In the context of ICI therapy, hepatitis, characterized by elevation of serum alanine transaminase and/or aspartate transaminase, is often initially clinically asymptomatic but can potentially lead to transient life-threatening liver dysfunction (35). It occurs in 19% (vs. 9% for grade 3-4) of patients treated with ICI-combination therapy, in 5% (vs. 2% for grade 3-4) of patients receiving anti-CTLA-4 and in 19% (vs. 9% for grade 3-4) of patients receiving anti-PD-(L)1 treatment (34). Although hepatic toxicity can occur with a delay of months to years, it typically manifests between 1 and 15 weeks after treatment (36). Recently, in a large population of melanoma patients treated with ipilimumab and/or nivolumab, especially in ICI-combination treatment caused higher rates of grade 3-4 liver toxicity with aminotransferase levels of 6.1% (for aspartate aminotransferase) and 8.3% (for alanine aminotransferase), and have been shown to lead frequently to ICI treatment discontinuation (44). Evidence for hepatitis can be found on US as a diffusely hypoechogenic liver parenchyma with periportal thickening and hyperechogenic dots, known as “starry sky pattern” sign ( Supplementary Figure 1 ), as well as gallbladder wall thickening (45). CT and MRI findings are often nonspecific and comparable to those of other causes of acute liver dysfunction, ranging from the absence of detectable abnormalities to hepatomegaly, heterogeneous parenchymal enhancement with areas of low attenuation, periportal/gallbladder edema (diffuse parenchymal hypoattenuation on CT or T2-weighted hyperintensity on MRI), and perihepatic ascites (36, 37, 42, 45). Interestingly, while other causes of diffuse liver disease might not be visualized on PET/CT, increased liver ¹⁸F-FDG avidity in the setting of ICI-induced hepatitis is often reported (46). However, its visualization using ¹⁸F-FDG PET/CT might be limited by physiological uptake or a reversal in liver-to-spleen ratio due to higher spleen uptake resulting from ICI-induced T-cell activation especially at an early stage (47). Some authors have suggested the use of a liver-to-blood pool standard uptake value (SUV) mean ratio to detect pathologic hepatic uptake and thus possible hepatitis compared to SUVmean alone because various parameters can influence SUV measurements (48, 49).

Cholecystitis and cholangitis are forms of hepatobiliary toxicity that are rarely associated with ICI therapy, and because of the small reported number of cases, it is difficult to estimate the actual incidence and causal relationship, if any, with immunotherapy (35, 50, 51). Despite its low incidence characteristic imaging features on US and CT for cholecystitis such as gallbladder distension and wall thickening, as well as inflammation of the pericholecystic tissue with increased ¹⁸F-FDG uptake on PET/CT, have been suggested (52). On MRI and MR cholangiopancreatography (MRCP), as the imaging modalities of choice in clinical practice, cholangitis is characterized by localized dilatation and diffuse non-obstructive hypertrophy and hyperenhancement of the extrahepatic bile duct wall with portions of biliary dilatation and narrowing ( Figure 2 ) (51, 52).

The diagnosis of pancreatitis requires the presence of at least two of the following three features: abdominal pain suggestive of pancreatitis, elevated amylase or lipase to more than three times of the upper normal limit, and characteristic imaging features (53) because ICI-related acute pancreatitis is relatively rare and patients with elevated amylase and lipase are often initially asymptomatic, only few cases have been reported (54). Radiologic findings on CT and MRI are similar to pancreatitis from other origin and include in the acute phase focal or diffuse pancreatic enlargement with decreased enhancement and peripancreatic fat stranding associated with edema and fluid collections without a focal lesion suspicious for metastasis (37, 42, 55). On ¹⁸F-FDG-PET/CT diffuse tracer uptake might be seen (56). After resolution of the clinical presentation, imaging might be characterized by parenchymal atrophy and loss of normal lobulations (55).

Acute kidney injury (AKI) is the most common renal toxicity in patients receiving ICI therapy (57). However, it is generally not a direct consequence of ICI’s toxicity, as it can be caused by various etiologies. Therefore, it is important to distinguish between AKI as an irAE and AKI induced by e.g. hypovolemia or acute tubular necrosis (57). Notwithstanding, the incidence of AKI after ICI treatment is reported to be 2.2% overall and 0.6% in severe cases requiring renal transplantation (58). AKI occurs more frequently in patients treated with ICI-combination therapy (4.9%) than with anti-CTLA-4 (2%) or anti-PD-(L)1 (1.4%-1.9%) monotherapy (58). The interval between ICI treatment initiation to AKI ranges from 21 to 245 days, and from 7 to 63 days between the last ICI treatment dose and onset of AKI (58) A ccase of ipilimumab-induced immune-related kidney failure was reported with bilateral renal enlargement visualized on CT and rapid resolution after steroid therapy (59). In addition, PET/CT shows increased ¹⁸F-FDG uptake in the renal cortex (60, 61). Moreover, diffuse or even segmental uptake in the renal parenchyma can be seen on ¹⁸F-FDG PET/CT especially in delayed imaging. However, the very high pelvicalyceal activity and low spatial resolution in older generation PET/CT scanners are clear limiting factors for accurate assessment of the kidneys in patients undergoing ICI treatments (42). Therefore, for patients with clinical suspicion of AKI and contraindication for biopsies, ¹⁸F-FDG PET/CT might provide some diagnostic clues (60). However, the specific imaging characteristics have not been defined yet and distinguishing between immune-related and non-immune-related AKI remains challenging.

Pneumonitis is a relatively common irAE that manifests with clinical symptoms ranging from mild dyspnea to potential lethal respiratory failure and is associated with lower patient survival (62). Pneumonitis occurs in approximately 1% of patients receiving anti-CTLA-4 therapy and in 4% of patients receiving anti-PD-(L)1 treatment, with around 1% of the cases being severe (34). In patients receiving ICI-combination treatment (nivolumab + ipilimumab or peptide vaccines), the incidence is significantly higher at 6.6% and 1.5% for severe cases, respectively (63). The median time to onset of clinical symptoms is 4.6 months in patients receiving ICI-monotherapy versus 2.7 months in patients receiving ICI-combination therapy (64, 65). Radiologic findings of ICI-related pneumonitis range from mild interstitial abnormalities to acute interstitial pneumonia and acute respiratory distress syndrome (66). The best imaging modality in this setting is CT. Based on the CT findings, irAE-related pneumonitides can be divided into five distinct phenotypes (37, 65):

Ground glass opacities (55%) and consolidations (32%) non-segmentally distributed in the dominant lung or bilaterally opposite the tumor have been shown the most frequently in ICI-treated non-small cell lung cancer patients according to a systematic review by Zhang et al. (67).

A case of progressive pleural effusion as rare irAE has been reported for instance in a non-small cell lung cancer patient treated with cisplatin, pemetrexed, and pembrolizumab (68) and as late toxicity in a primary lung adenocarcinoma patient following 94 cycles of nivolumab (69). Importantly, CT scans are also of interest for ruling out differential diagnoses such as pulmonary embolism.

On PET/CT an interstitial pneumonia type pattern characterized by non-specific moderate to intense ¹⁸F-FDG uptake might be seen (70). However, one major diagnostic challenge is to distinguish infectious diseases from tumor lesions, e.g., nodular aspects that mimic tumor recurrence, whereas an underlying disease, such as chronic obstructive pulmonary disease, may further complicate the final diagnosis (71).

In terms of clinical management, corticosteroids are recommended as primary therapy approach based on severeness of the case and clinical expertise (67). In addition, for patients with grade 3-4 ICI-induced pneumonitis, ICI-treatment should be discontinued immediately and permanently. Clinical improvement, especially in low-grade disease is usually observed within 48-72 hours of corticosteroid use. Patients with grade 2 pneumonitis, who resolved symptoms show the highest overall survival (86%) compared with grade 3 or 4 pneumonitis (36% or 43%, respectively) (67).

Immunotherapy-related sarcoid-like reactions are often asymptomatic and appear in 5–7% of patients (37, 72). They might be related to the involvement of primary and secondary systemic lymphoid organs in the systemic antitumor response required for effective ICI treatment (37, 72, 73). In general, the formation of sarcoid-like granulomas occurs most frequently in lymph nodes (71%), lungs (60%), and skin (55%) and can be easily confused with disease progression or tumor recurrence (7, 74). The median time between initiation of ICI treatment and the development of sarcoid-like reactions is 14 weeks (37, 75). During the course of ICI therapy, metabolic changes in lymphoid organs could be monitored using ¹⁸F-FDG-PET/CT. Indeed, an increase in immune cells populations and their higher growth rate leads to higher energetic requirements often translating in high avidity for ¹⁸F-FDG (76, 77). Imaging findings include a new bilateral symmetric mediastinal and hilar lymphadenopathy resembling sarcoidosis in up to 10% of the patients (78), often high ¹⁸F-FDG avidity on PET/CT, and an association with (subpleural) perilymphatic distribution of micronodules without suspicion of intercurrent infection or new metastasis ( Figure 4 ) (37, 79). It is critical to recognize these sarcoid-like irAEs as a classic response pattern to immunotherapy (initial increase of the tumor burden unconfirmed at the next imaging follow-up) in order to distinguish it from true progression or pseudoprogression (80). In cases where the diagnosis on imaging is unclear, an assessment of angiotensin converting enzyme serum levels can be performed, as elevated levels have been associated with ICI-induced sarcoidosis-like reactions (75). If true tumor progression is still suspected after this, a targeted biopsy should be strongly considered for definitive diagnosis (75).

Cardiac toxicities associated with ICI treatment are relatively rare. Myocarditis, as the most common one, occurs in 0.1% to 1% of patients with symptoms such as dyspnea (49%), weakness (25%), chest pain (17%), syncope (9%), fever (6%), and cough (4%) (20, 21, 81). In most of these cases, the onset is shortly after initiation of the ICI therapy, and because of the high mortality rate of 50%, it is of the utmost importance to make the diagnosis and start the appropriate treatment as early as possible (20, 21, 82). Besides clinical features, laboratory markers and electrocardiogram changes, non-invasive imaging modalities, especially cardiac MRI (CMRI) has become more and more important in the diagnostic workup, to reduce the necessity of invasive biopsies as the current diagnostic gold standard (83, 84). In clinical practice, transthoracic echocardiography (TTE) is the first imaging modality that should be performed if acute myocarditis is suspected. Suggestive TTE findings include abnormalities of the segmental wall motion, increased thickness of the left ventricular wall, global hypokinesia (fulminant myocarditis), and pericardial effusion (84). However, a recent review of 88 ICI-induced myocarditis cases showed normal morphological TEE findings in 23% and normal left ventricular ejection fraction in 32.5% (81). Regarding CMRI, at least one criterion on T2-based (regional or global increase in myocardial relaxation time or increased signal intensity) with at least one criterion on T1-weighted imaging (increase in myocardial T1, extracellular volume, late gadolinium enhancement) should be analyzed for sufficient diagnostic accuracy according to the recently updated Lake Louise criteria (85). In 48% of cases late gadolinium enhancement (LGE) predominantly distributed in the anteroseptal, inferoseptal, inferior, and inferolateral segments (atypical localizations possible), and in 28% of the cases myocardial oedema in T2-weighted short-tau inversion recovery (STIR) is described (83, 86). However, these characteristics are limited due to their low specificity. In a study of an international registry of patients with ICI-associated myocarditis (n=103), only 48% of patients with ICI-induced myocarditis had LGE when compared to 90% of patients with other causes of myocarditis (83). Herby it is important to note that CMRI assessment >4 days after admission showed significantly more positive LGE findings than if CMRI was performed earlier (72.0% vs 21.6%, p<0.001) (83). Moreover, LGE was not associated with clinical symptoms, patient outcomes, ECG or echocardiographic findings. Finally, nuclear medicine findings might provide important clues for the diagnostic of immune-related acute myocarditis ( Figure 5 ). Interestingly, ¹⁸F-FDG-PET/CT has a very limited role in this setting as demonstrated in a recent study of 61 patients with suspicion of ICI-related myocarditis where its sensitivity was below 30% (87). However, other PET tracers have been proven useful in this context, such as ⁶⁸Ga-DOTATOC which showed high sensitivity for the early detection of pathological myocardial uptake in a small population of patients (n=9) with clinical suspicion of ICI-related myocarditis (88). A pathological diffuse tracer uptake in the myocardium was the most frequent pattern detected. Interestingly, ⁶⁸Ga-DOTATOC PET/CT showed a good correlation with elevated serum cardiac troponin I and immune correlates such as inflammatory cytokines (IL-6) and chemokines (CXCL9, CXCL10 and CXCL13) by contrast evocative lesions for myocarditis were only seen in 3 out of the 8 patients that had a CMRI (88).

Pericarditis is reported to be the second most common immune-related cardiotoxicity, although data is lacking regarding its exact incidence (82). The median onset of pericardial disease is estimated to be 30 days (82). Symptoms include shortness of breath, pericardial pain without pericardial effusion or jugular vein congestion, and cardiogenic shock with cardiac tamponade due to pericardial effusion, resulting in a high mortality rate of 21% (82, 89). Moreover, pericardial toxicity can occur alone or in combination with ICI-associated myocarditis (myopericarditis) (89). The diagnostic work-up includes detailed physical examination, electrocardiogram, echocardiogram, CMRI and cardiac PET/CT (89, 90). On CMRI, ICI-related pericardial disease demonstrates focal myocardial LGE in the mid-lateral wall and mild LGE of the pericardium along the lateral wall in cases suggestive of myopericarditis (90).

Regarding peripheral neuromuscular toxicities, myositis is the most common syndrome. While being the most prevalent in anti-PD(L)-1 therapy, it occurs in approximately 0.4-3% of ICI-treated patients ( Figure 6 ) (91–93). The median time of ICI-administration to myositis symptom development ranges from 5 to 87 days (94). Interestingly, the clinical manifestation from immunotherapy-related myositis differs markedly from that of idiopathic and paraneoplastic inflammatory myopathies such as dermatomyositis and polymyositis. Progressive symptom development, as well as oculomotor and axial muscle involvement are uncommon, but have been reported. Bulbar symptoms, such as dyspnea, dysarthria, and dysphonia have been described (95, 96). However, sudden onset of stable myalgia with or without elevated creatine kinase is the most common symptom of immune-related myositis (96). The differential diagnosis to myastenia gravis is sometimes challenging, since on one side, myastenia gravis is often associated with optical myositis and on the other side, acetylcholine receptor binding antibodies can occasionally be detected in optical myositis in the absence of myasthenia gravis (97, 98). On brain MRI, immunotherapy-related myositis is characterized by fat-suppressed T1/T2-weighted intramuscular hyperintensity with or without gadolinium enhancement (96, 99). The ocular phenotype presents contrast-enhanced orbital edema as well as abnormal enhancement and enlargement of the extraocular muscles ( Figure 7 ). Moreover, PET/CT with increased muscular ¹⁸F-FDG uptake can support the diagnosis and help to estimate the severity by assessing how many muscle groups are affected (96, 99). Pathological muscle uptake suggestive of myositis could also be detected using ⁶⁸Ga-DOTATOC PET/CT as shown in 5 out of the 6 patients that presented with myositis concomitant to an ICI-related myocarditis (88). In addition, rheumatological disorders are also frequent irAEs during the course of ICI treatment and ¹⁸F-FDG-PET/CT could be useful for the detecting and assessing the severity of the inflammation associated with those events in particular for arthritis affecting several articulations but also tenosynovitis or polymyalgia rheumatica (41).

In contrast to peripheral nervous toxicities, irAE of the central nervous system such as encephalitis, aseptic meningitis, vasculitis, cranial neuropathies, and myelitis are uncommon (100).

Although in recent years an increasing number of immune-related encephalitis have been described and may occur with each treatment cycle, it remains a rare immune-related toxicity with an incidence of 0.1–0.2% (94, 100). These cases present a wide range of potential life-threatening symptoms, including confusion, agitation, fever, headache, fatigue, short-term memory impairment, neck stiffness, behavioral changes, and psychiatric symptoms (101, 102). The diagnostic workup usually includes brain MRI, lumbar puncture, paraneoplastic autoantibodies, electroencephalography, and laboratories, notably to rule out infectious agents related diseases. MRI, particularly T2-weighted and/or fluid-attenuated inversion recovery (FLAIR), reveals encephalitis features such as ill-defined uni- or bilateral hyperintense signals in the limbic cortex, the cerebellum, basal ganglia or scattered in the gray or white matter, with or without enhancement corresponding to zones of inflammatory infiltrates and epileptogenic activity (93, 103, 104), some being associated with auto-antibodies (105). Multifocal lesions involving the white matter, optical nerve, and spinal cord, which mimic demyelinating diseases, have also been described (106, 107). The physiological high ¹⁸F-FDG uptake of the brain limits somehow the irAEs assessment using ¹⁸F-FDG-PET/CT (41). However, there is evidence of the utility of ¹⁸F-FDG-PET/CT, showing increased or decreased metabolic activity, to detect ICI-induced encephalitis earlier than standard diagnostic approaches (108). A recent study in patients with autoimmune encephalitis, which shares many similarities with ICI-related encephalitis, described 6 cases with metabolic abnormalities on ¹⁸F-FDG-PET/CT with normal MRI (n=2), lumbar puncture (n=3), and electroencephalography (n=2) findings (109). Finally, cases of posterior reversible encephalopathy syndrome have been occasionally reported following ICI administration alone or in combination with chemotherapy (110–112).

Aseptic meningitis is present in <0.1% of ICI-treated patients overall (93). This condition is more commonly associated with anti-CTLA-4 and ICI-combination treatments (93, 100). Moreover, the time to clinical disease onset is short, with a delay of 9 days from the first dose of immunotherapy (100). Aseptic meningitis is characterized by the subacute onset of nonspecific symptoms such as headache, neck stiffness, photophobia, low-grade fever, and nausea, and must be distinguished from infectious or carcinomatous causes of meningitis (93). In 42% of the patients, brain MRI shows diffuse leptomeningeal enhancement with or without parenchymal abnormalities as a nonspecific sign of inflammation and is consistent with the presence of lymphocytic or neutrophil pleocytosis, while overlapping with findings of immune-induced meningoencephalitis (113, 114)( Figure 8 ). However, 46% of brain MRI findings are normal in ICI-induced aseptic meningitis (113). However, this even underlines the importance of imaging to rule out differential diagnosis such as (ischemic) stroke, infection, and brain metastasis.

In recent years, there is increasing evidence for ICI-associated central nervous system vasculitis. However, the exact frequency (currently estimated to be <0.01%), timing and association with a specific type of immunotherapy is still unclear (93). Commonly reported types of vasculitis are large vessel vasculitis (giant cell arteritis and isolated aortitis) and vasculitis of the nervous system (primary angiitis of the central nervous system, PANCS, and isolated vasculitis of the peripheral nervous system) (93, 115). With a median time of 3 months from the initiation of ICI treatment to their development, symptoms are often unspecific and include headaches (60%), altered cognitive status (50%), and focal neurologic deficits. Most commonly, they are mild, and no fatalities related to vasculitis have been observed (93, 115). Although considered to be the gold standard for diagnosis, biopsy of the brain and/or spinal cord showing segmental inflammatory infiltration leading to blood vessel walls thickening and stenosis, resulting in decreased blood flow or even secondary to hemorrhagic vessel rupture, has only a sensitivity of 53% (116). However, this sensitivity can be increased to more than 80% by identifying focal lesions previously on neurologic imaging techniques (117). Brain MRI is altered in more than 90% of patients with PANCS, showing (nonspecific) signs of microangiopathy, hemorrhage, or ischemic infarction, as well as multifocal bilateral T2- weighted, FLAIR and diffusion-weighted sequence abnormalities in the cortical-subcortical area (118). However, the occasional presence of solid lesions and gadolinium enhancement of leptomeninges complicate the distinction to tumors and abscesses and requires additional imaging modalities such as CT angiography, high-resolution contrast-enhanced MRI, or ¹⁸F-FDG-PET/CT to detect vascular inflammatory activity (119).

Endocrinopathies are observed in up to 10% of patients treated with anti-CTLA-4 and in 4-14% of patients treated with anti-PD-1 therapy (120, 121).

Hypophysitis occurs in 4.5% (0.8% severe cases) of the patients treated with anti-CTLA-4, whereas it is reported in less than 1% (0.1%) of patients with anti-PD-(L)1 treatment (122). The clinical features of pituitary dysfunction can be nonspecific and include fatigue, headache, or weakness with additional symptoms related to hypopituitarism (122, 123). The median time to symptom onset ranges between 11 weeks (ipilimumab), 17 weeks (combination of impilimumab and nivolumab), 22 weeks (nivolumab), and 26 weeks (pembrolizumab) (124). Since pituitary inflammation can be caused by ICI therapy as well as by pituitary metastasis and adenomas, MRI and ¹⁸F-FDG-PET/CT are playing a crucial role in distinguishing these diseases as they often show imaging findings of immune-related hypophysitis before the appearance of symptoms (18, 42, 45). Contrast-enhanced MRI of immune-related hypophysitis shows enhancement of the posterior portion of the pituitary gland in 89% of the patients, whereas the enhancement is homogeneous in 63.3% (vs. heterogeneous enhancement, 36.7%) (16, 125) ( Figure 9 ). This pattern is important for distinguishing this toxicity from pituitary metastasis, as differential diagnosis, which show heterogeneous enhancement in the vast majority of cases (82.6%) (16). Moreover, thickening of the pituitary stalk has been identified in 29/49 (59.2%) cases of hypophysitis and only in 16/58 (27.6%) cases with pituitary metastasis (16). On PET/CT, immune-related hypophysitis shows ¹⁸F-FDG-avid pituitary gland often enlarged but without mass effect on the optic chiasm and with thickening of the infundibulum (125, 126). A recent study in 162 advanced melanoma patients who received ipilimumab/nivolumab combination therapy showed that ¹⁸F-FDG-PET/CT was able to predict the appearance of hypophysitis with high positive (86%) and negative (87%) predictive values (127).

ICI-induced thyroid dysfunction is often clinically asymptomatic and transient, and identified by blood tests as mild hypo- or hyperthyroidism associated with elevated anti-thyroid peroxidase and/or anti-thyroglobulin antibodies (128). In terms of frequency, hypothyroidism is more common, affecting 15% of patients receiving ICI-combination therapy, 3% of patients receiving anti-CTLA-4, and 8% of patients receiving anti-PD-(L)1 therapy (34). Hyperthyroidism is observed in only 4% of patients treated with anti-CTLA-4 and 5% of patients treated with anti-PD-(L)1 molecules (34). In few cases, ICI therapy did lead to Graves’ disease (129). Immune-related thyroiditis, which usually occurs within 5 to 10 weeks following treatment, is mostly mild and CTCAE grade ≥3 is rarely observed (34, 130). US is the imaging modality of choice and new enlargement of the thyroid gland with heterogeneous hypoechoic parenchyma, (pseudo)nodular pattern, and increased vascularity on color Doppler is commonly observed (37, 42). CT findings are unspecific as they present a new enlargement of the thyroid gland associated with a heterogeneous parenchymal enhancement (37, 42). Still, thyroiditis remains frequently an incidental finding on ¹⁸F-FDG-PET/CT with a diffuse increased uptake of the thyroid gland ( Figure 10 ) (37, 42).

Compared with the more common secondary adrenal insufficiency caused by pituitary dysfunction, primary adrenal insufficiency, in which the adrenal glands are directly damaged due to ICI therapy, has been rarely described (131). Adrenal insufficiency is estimated to occur in 5% of patients treated with ICI-combination therapy and in 1% of patients treated with anti-CTLA-4 or anti-PD-(L)1 mAbs, whereas CTCAE grade ≥3 is rarely reported (34, 132). Disease onset after initiation of ICI treatment is in between 9 weeks (ipilimumab), 3.3 month (pembrolizumab) and 5 months (nivolumab) (121, 133). Symptoms are characterized by electrolyte abnormalities, dehydration and altered mental status. Life-threatening adrenal crisis with vasodilatator shock and hypotension, requiring permanent steroid replacement therapy, was reported following nivolumab therapy (134). Therefore, rapid diagnosis and close monitoring are required. On CT and MRI, adrenal glands show bilateral, symmetrical, and smooth enlargement, while uniform mild hypermetabolism is seen on ¹⁸F-FDG PET/CT (42, 135).