Key points
- The emergence of cancer immunotherapy has revolutionized cancer treatment but is associated with serious immune-related adverse effects (IRAEs)
- Cytotoxic T-lymphocyte antigen 4 (CTLA4)-targeted immunotherapy is associated with increased susceptibility to hypophysitis and primary thyroid dysfunction
- Programmed cell death protein 1 (PD1)-targeted immunotherapy is associated with primary thyroid dysfunction and type 1 diabetes mellitus
- CTLA4–PD1 combination therapy has an elevated incidence of hypothyroidism and possibly incidence rates of hypophysitis similar to those with monotherapy with CTLA4 antibodies
- IRAEs might be associated with improved clinical response of tumours to immunotherapy, but further studies are needed to evaluate this possible effect
Main
The increased understanding of the human immune system and the emergence of immune modulation techniques have led to a new era in cancer therapy, and the idea of using our own biology to treat cancer is a revolutionary area of oncology. To ensure that the immune system does not harm the host when reacting to a foreign antigen, humans have evolved immune checkpoint proteins and mechanisms to quickly halt an immune response. However, in the case of cancer, malignant cells have developed many mechanisms to evade the human immune system [1,2], including the ability to limit immune responses through such immune checkpoints [3]. New cancer therapies have made use of the accumulating knowledge regarding immune regulation and immune system checkpoints; for example, cytotoxic T-lymphocyte antigen 4 (CTLA4) and the programmed cell death protein 1 (PD1) pathway.
In resting T cells, CTLA4 resides intracellularly but is translocated to the plasma membrane shortly after T cell activation [4]. In an active immune response, CD28 on the T cell surface binds to the B7 co-stimulatory ligand on antigen-presenting cells to provide the second signal that allows the T cell to mature [4]. CTLA4 binds with high affinity to B7 and can compete with CD28 to further inhibit T cell activity [5]. This process prevents the second signal that supports T cell activation and effectively stops the T cell from maintaining an immune response [6] (Fig. 1). Monoclonal antibodies that target CTLA4, such as ipilimumab, have demonstrated efficacy in cancer treatment [7,8] (Fig. 1). The binding of these antibodies to CTLA4 results in prevention of B7 binding; with B7 now accessible, CD28 enables upregulation of T cell activity [4]. CD28-initiated downstream activation of mitogen-activated protein kinase results in the formation of activator protein 1 (AP-1) complex [9]; in conjunction with T cell receptor-mediated nuclear factor of activated T cells signal, the AP-1 complex induces IL-2 cytokines, which mediate T cell growth [9]. With CTLA4 blocked, activated T cells proliferate and achieve a persistent state of activation, which enables the targeting of otherwise poorly immunogenic tumour antigens to cancer cells [10].
PD1 is an immune cell-specific surface receptor [11,12], and ligands for PD1 (PDL1 and PDL2) are associated proteins that are found on antigen-presenting cells and cancer cells [13–16]. When bound to a ligand, PD1 lowers the threshold for apoptosis, induces anergy via blunted T cell receptor signalling and generally leads to T cell depletion [5,17] (Fig. 1). In certain tumour cells, upregulation of PDL1 expression has been observed, which leads to increased inhibition of T cell activity in favour of tumour cell survival [18,19]. A monoclonal antibody against PD1 can block this pathway (that is, a PD1–PDL1 interaction) and result in upregulation of immune response and inhibition of tumour growth [20–23] (Fig. 1).
Suppressing these immune checkpoints results in immune-mediated antitumour activity in mouse models and clinical trials7,8,15,20,24,25,26. Specifically, suppression of the CTLA4 and PD1 pathway enables the expansion of tumour-specific T cells [5,20]. However, immunotherapy has also led to immune-related adverse events (IRAEs) [27,28], which can range from mild to fatal, depending on the organ system and severity [27]. Although endocrinopathies are not among the most common IRAEs reported, they can be particularly severe and, consequently, must be carefully monitored for during treatment with immunotherapeutic agents [29].
The two main endocrinopathies observed with checkpoint blockade treatments include hypophysitis (typically induced by CTLA4 antibodies) and primary hyperthyroidism or hypothyroidism (induced by antibodies against PD1, PDL1 or CTLA4) [30–32]. The precise mechanisms remain unclear; however, possible pathophysiologies are currently being evaluated in mouse models [33]. CTLA4 and PD1 monoclonal antibodies target different mechanisms [5]; whereas CTLA4 is involved in initial T cell deactivation, PD1 targets the modulatory phase of the immune response [34,35], which might, in part, explain the differences in IRAEs between the two therapies.
Interestingly, a correlation seems to exist between overall patient survival and the incidence and severity of IRAEs [36,37]. This trend might be due to the monitoring of patients for a longer period of time and the bias resulting from extended duration of symptomatic observation [38,39]. However, the correlation could also be the result of autoimmunity being indicative of a nonspecifically overactive immune system resulting in increased antitumour efficacy [40]. This notion is supported by a decrease in cancer-specific mortality in patients who also experience IRAEs, including endocrine IRAEs [36,40]. With the increased clinical application of immunotherapeutics, understanding the prevalence, detection and management of IRAEs in the patient population is important.
In this Review, we focus on the endocrinopathies that are related to four immunotherapeutic agents for which the largest amount of safety data is available: ipilimumab (CTLA4), tremelimumab (CTLA4), nivolumab (PD1) and pembrolizumab (PD1). We also briefly discuss the antibodies that block PDL1, which are in clinical development and have toxicities that generally resemble those of PD1 blocking antibodies [26].
CTLA4 antibodies
Ipilimumab (also known as MDX-010) was first shown to be efficacious in two different phase III trials in patients with metastatic melanoma [7,8] and was subsequently approved by the FDA for treatment of unresectable or metastatic melanoma [41]. Tremelimumab (first known as CP-675,206) works through a mechanism that is similar to that of ipilimumab to block CTLA4 (Refs 42–44).
The reported adverse effects of ipilimumab include dermatitis, enterocolitis, hepatitis and uveitis, some of which have been classified as severe grade 3–4 effects, principally owing to colitis or hepatitis [45]. Endocrinopathies are of special interest, as they are considered rare in the general population, with idiopathic autoimmune hypophysitis only observed in 1 in 9 million individuals [46,47]. Pituitary dysfunction is among the most commonly reported endocrinopathies (mean 9.1%; Table 1) that are associated with ipilimumab; other reported endocrinopathies include primary hypothyroidism, primary hyperthyroidism and primary adrenal insufficiency [31,32].
Hypophysitis
The incidence of hypophysitis related to CTLA4 antibody therapy has been reported to vary from 0.4% to 17% in patients treated with this therapy [31]. This wide range in incidence has been associated with differences in hormonal monitoring, drug dose and improved recognition of anti-CTLA4-related hypophysitis, with low incidences being reported in initial studies [29,36]. Two reports specifically studying anti-CTLA4-related hypophysitis and other endocrine dysfunctions during ipilimumab therapy noted an 8–13% incidence of hypophysitis [32,36,48,49], and an 8.5–9.0% incidence of hypophysitis with ipilimumab combined with the PD1 monoclonal antibody nivolumab has also been reported [32,50]. In our own analysis of clinical trials and retrospective reviews, we found an overall incidence of hypophysitis in 9.1% of patients (184 of 2,017; Table 1); the incidence of secondary adrenal insufficiency, secondary hypothyroidism and secondary hypogonadism were 6.1%, 7.6% and 7.5%, respectively. Importantly, we only included instances with a clear secondary aetiology for adrenal insufficiency and hypothyroidism. Overall, many studies lacked specific endocrine data, and further studies are therefore needed with in-depth hormonal investigations.
The incidence of anti-CTLA4-related hypophysitis is associated with the dose of therapy received51, although there have been conflicting reports regarding this finding [29,32,36,49,52,53,54]. In addition, other investigators have reported that hypophysitis incidence was not significantly different among cohorts who received 3 mg/kg and 10 mg/kg (Refs 32,52). Some investigators argue that cumulative treatment effects might influence the timing of hypophysitis after treatment initiation, as many patients do not become symptomatic until ∼11 weeks after the first dose [7,55,56]. The time for the onset of hypophysitis symptoms can range from 6 weeks to 12 weeks after CTLA4 antibody treatment initiation [29,31,32,36,49], but patients can present as early as week 4 and as late as week 16 (Refs 56,57).
Hypophysitis related to CTLA4 antibody therapy may occur more commonly in men than in women. In one report, the prevalence of anti-CTLA4-related hypophysitis was 15.6% in men and 3.6% in women (P = 0.02; OR 4.73; 95% CI 1.27–30.79) [36]. This finding is in contrast to idiopathic autoimmune hypophysitis, which is more common in women (men to women ratio 1:3) [30,46,58,59]. This phenomenon has been suggested to be the result of the increased prevalence of men with melanoma in these trials32. In terms of age, one study reported a statistically significant older population (mean age of 68.2 ± 2.4 years compared with 59.9 ± 1.0; P = 0.005) being affected by ipilimumab-induced hypophysitis [36].
Headache, fatigue and/or muscle weakness seem to be the most common (in 89% of those diagnosed with hypophysitis) presenting symptoms of hypophysitis related to anti-CTLA4 therapy [32]. However, these symptoms are nonspecific and could be misattributed to general symptoms related to cancer. Nausea, anorexia, weight loss, visual changes, alterations in mental status, temperature intolerance and arthralgias are also reported, but less frequently (10.5–21.1%)29,32,36. Low levels of sodium (range from 113 mEq/l to 134 mEq/l) have also been reported in patients with anti-CTLA4-related hypophysitis, with some studies reporting hyponatraemia in 47–56% of patients [32,36], although this was not reported in all studies [7,56,60]. Morbidity attributed to anti-CTLA4-related hypophysitis is thought to be predominantly related to secondary adrenal insufficiency [27], which might be life threatening if not treated. Symptoms of adrenal crisis include hypotension, dehydration and electrolyte imbalance, and require immediate attention [27]. Notably, the symptoms of hypophysitis can improve in as quickly as days after starting treatment with corticosteroids [32].
Adrenocorticotropic hormone (ACTH) and/or TSH deficiency are the most common pituitary hormone abnormalities that have been described in patients with anti-CTLA4-related hypophysitis [32,36,61,62]. These anterior pituitary hormone deficiencies are more prevalent than diabetes insipidus (with only a single case report of anti-CTLA4-associated diabetes insipidus) or posterior pituitary deficiency [55,56]. Hypogonadotropic hypogonadism has also been reported, although this might be confounded by the existence of hypogonadism induced by severe illnesses, such as sepsis [29]. Low levels of insulin-like growth factor 1 (IGF1) might also be present, but the growth hormone axis is less often evaluated, as treatment with growth hormone replacement is given to patients with active malignancy [29,36]. Prolactin has been described as being both elevated and low in patients with anti-CTLA4-related hypophysitis [31,32,59].
Adrenal insufficiency associated with anti-CTLA4-related hypophysitis is usually permanent [31,32,36,49], and these patients typically need life-long steroid replacement after developing this complication [49]. The recovery of secondary hypothyroidism can occur, but the frequency of this has been reported to vary from 6% to 64% [31,36,49,56]. Gonadal axis recovery has also been noted to vary from 11% to 57% [31,36,55,57,59]. Initial assessment of thyrotroph and gonadotroph function is complicated in a patient with cancer who undergoes cancer therapy, which can result in thyroid and gonadal laboratory values that are similar to the values that are observed in hypopituitarism [32]; furthermore, recovery from illness can normalize these hormone levels (for example, sick euthyroid syndrome or sickness-induced hypogonadism) [32]. Consequently, determining whether recovery from thyroid or gonadal hormone deficiency was due to improvement in hypophysitis or to simple recovery from the underlying illness is difficult.
Mild-to-moderate diffuse enlargement of the pituitary gland, with either homogenous or heterogeneous enhancement after contrast administration, is typically seen on sellar MRI in patients with anti-CTLA4-related hypophysitis [31,36]. The pituitary stalk might thicken, and, although uncommon, pituitary enlargement could result in a mass effect on the optic apparatus [36,49]. In a retrospective review of MRI data, relative pituitary enlargement was seen to precede the clinical diagnosis of anti-CTLA4-related hypophysitis [36]. This observation is further reiterated by the finding that the median onset time of pituitary enlargement is 1 week before any biochemical evidence of hypopituitarism [49]. The pituitary gland is thought to decrease in size over ∼4–12 weeks, and subsequent atrophy of the gland can be seen [31,36,55,57,63,64]. However, importantly, a normal MRI does not rule out hypophysitis, and management should be based on clinical presentation and evaluation of pituitary hormone levels45.
Mechanism
The mechanism of CTLA4 antibody-mediated endocrinopathy remains unclear. Autoantibodies targeting the pituitary have been described in patients (seven of seven) with ipilimumab-induced hypophysitis; these antibodies were not found in an ipilimumab-treated cohort (13 of 13) who did not have pituitary abnormalities33. Furthermore, in ipilimumab-induced hypopituitarism, TSH-targeting antibodies were identified in all patients (n = 7); other endocrine cell-targeting antibodies were also identified (FSH-secreting cells in five patients and ACTH-secreting cells in three patients)33.
Investigators have suggested that hypophysitis is caused by complement activation from antibody immunity developed against the pituitary gland [33]. Specifically, a type II hypersensitivity reaction to ectopic CTLA4 protein expressed on pituitary cells is thought to result in damage to the pituitary [65] (Fig. 2). Interestingly, patients treated with PD1 and/or PDL1 IgG4 antibodies, instead of IgG1 (which is used in ipilimumab), rarely developed pituitary damage [26,66,67]. This finding led researchers to hypothesize that IgG1, which activates the classic complement pathway, might be a possible mechanism of anti-CTLA4-related hypophysitis33. In support of this hypothesis, the occurrence of hypophysitis in patients receiving ipilimumab (IgG1) is notably elevated (9.1%) compared with its occurrence in patients receiving tremelimumab [68,69,70,71] (IgG2b; 1.3%; Tables 1,2). However, on the basis of our review of the literature, tremelimumab (n = 773) was not clinically evaluated as robustly as ipilimumab (n = 2,938), and the direct comparison of IRAE incidences might not provide a definitive understanding of their toxicity.
Monitoring and management
Patients should be informed of the symptoms of anti-CTLA4-related hypophysitis, which can present between treatment and clinical visits. Baseline and follow-up thyroid function tests (TFTs) have been recommended after treatment with anti-CTLA4 therapy [38]; however, in our experience, screening for secondary adrenal insufficiency is often not a component of routine monitoring. Given that adrenal insufficiency can be life threatening, and given the relatively high incidence of hypophysitis in patients treated with CTLA4 antibody therapy, routine monitoring with early-morning ACTH and cortisol levels at baseline and during treatment should be considered [27]. While receiving CTLA4 therapy, these tests can be performed monthly for the first 6 months, given that anti-CTLA4-related hypophysitis tends to occur early in the course of treatment. If the tests are normal and the patient is asymptomatic, testing can be done every 3 months for the next 6–12 months followed by further tests every 6 months for the following 2 years. When patients have symptoms or signs of hypophysitis or hypopituitarism, they should have a prompt evaluation for these complications, which includes assessing the levels of early-morning ACTH, cortisol, TSH and free T4. If early-morning readings are not feasible or an urgent assessment is needed, samples can be taken at any time of the day. A very low random cortisol and ACTH level might be helpful in diagnosing secondary adrenal insufficiency. In the acute phase of pituitary damage, the adrenal glands might respond to ACTH stimulation normally, because the adrenal glands have not yet atrophied from the chronic lack of ACTH stimulation [72]. Consequently, a cosyntropin stimulation test is not as useful in diagnosing early secondary adrenal insufficiency. In patients with hypophysitis or hypopituitarism, the levels of gonadotropins and sex hormones should also be assessed. In those with secondary hypogonadism, prolactin levels can be measured.
High-dose steroids can be used for those patients with critical illness, either related to hypophysitis or hypopituitarism, significant hyponatraemia, severe headache, visual abnormalities or significant pituitary enlargement that abuts or has mass effect on the optic apparatus. Glucocorticoid treatment can decrease pituitary size gradually with symptom relief [55,57,59]. However, in a retrospective study, high-dose steroids did not seem to reverse hypopituitarism, and the investigators suggest that secondary hormonal abnormalities should be treated instead of the hypopituitarism [49]. By contrast, for idiopathic lymphocytic hypophysitis, spontaneous recovery of pituitary function, as well as recovery after high-dose steroids, has been described in some patients [73,74]. Low doses of glucocorticoids (for example, 15–25 mg of hydrocortisone in split doses or an equivalent dose of prednisone) can alleviate fatigue and headache, and treat those with adrenal insufficiency [54]. These regimens are also considered when low doses of steroids are needed for the patient to be eligible for clinical trials with immunotherapy drugs, as high-dose steroids can be part of the exclusion criteria given their possible immunomodulatory effect. Fortunately, the anticancer effects of CTLA4 immunotherapy do not seem to be influenced by treatment of anti-CTLA4-related hypophysitis with glucocorticoids [53,75–78].
Levothyroxine can be used to treat secondary hypothyroidism; but glucocorticoid deficiency should be first treated to avoid any potential adrenal crisis that can be precipitated by replacing thyroid hormone first [79]. Patients with central hypothyroidism or hypoadrenalism also often need long-term hormone replacement therapy [27,31,38,60,80]. Hyponatraemia is typically short-lived and improves after adrenal and thyroid hormone replacement [81]. Testosterone might be used in patients who develop hypogonadotropic hypogonadism, but the treatment should be consistent with the current Endocrine Society guidelines [82]. Oestrogen replacement can also be considered in premenopausal women who have secondary hypogonadism [83].
Primary thyroid dysfunction
Distinguishing primary thyroid dysfunction (that is, related to thyroid gland dysfunction) from thyroid dysfunction that is secondary to hypophysitis-related pituitary dysfunction is important for treatment [27]. Elevated levels of TSH with low-to-normal levels of free T4 or T3 are seen in primary hypothyroidism84. Low-to-mid-normal levels of TSH with low free T4 suggest hypothyroidism secondary to pituitary dysfunction [40]. However, tests for levels of T3 in patients with any acute illnesses can be inaccurate [84]. Other investigators have reported that the best assessment of primary thyroid dysfunction was by the measurement of TSH levels [40]. In thyroiditis, primary hypothyroidism (high TSH and/or low free T4) might be preceded by a transient hyperthyroidism (low TSH, elevated free T4 and/or T3) [85]. Unless thyroid dysfunction is subclinical and detected only via laboratory tests, hypothyroidism can be recognized by the presence of symptoms such as fatigue, muscle weakness, cold intolerance and bradycardia [85].
In clinical trials in which the patient received ipilimumab, the incidence of secondary hypothyroidism was 7.6% (4.3–11.0%), with primary hypothyroidism reported in 5.6% of patients (5.2–5.9%) [7,36,62,86–94] (Table 1). However, several of these trials did not include detailed information regarding aetiology, diagnostic test values or management of the hypothyroidism. For example, different trials reported unspecified hypothyroidism occurring in as few as 1.5% of patients and as high as 8.8% [7,36,62,86–88] (Table 1). The high occurrence of 8.8% might highlight the fact that this study enrolled patients who were free of radiographically detectable melanoma after surgery and received high-dose (10 mg/kg) ipilimumab88. The time to presentation of primary hypothyroidism after treatment with ipilimumab was not clearly defined. In a retrospective review, 154 patients had a normal baseline TSH before ipilimumab therapy, two developed transient thyrotoxicosis during treatment, followed by primary hypothyroidism, and six had newly elevated TSH levels during treatment or immediately after the conclusion of therapy [36]. However, in this study, data regarding clinical presentation, thyroid autoantibody and thyroid ultrasonography were not reported [36].
In a retrospective review of clinical trials using ipilimumab, primary hypothyroidism was identified in nine patients (who did not have concomitant PD1 blockade) [32]. In these patients, hypothyroidism presentation ranged from as early as 5 months to 3 years, but thyroid autoantibody data was not presented. The most common presenting symptom was fatigue, which improved with thyroid hormone replacement; three patients also developed thyroiditis while receiving ipilimumab [32]. The prevalence of subclinical primary hypothyroidism was best characterized in a retrospective study of 137 patients who were enrolled in the ipilimumab expanded access programme in which levels of TSH and free T4 were evaluated at baseline and during follow-up. Of these patients, six (4%) developed subclinical hypothyroidism, defined as a TSH level between 5 mIU/ml and 10 mIU/ml with normal levels of free T4 (Ref. 32). Two additional cases of ipilimumab-induced hypothyroidism (out of 27 patients) were also reported in a cohort of patients from Italy; one patient transitioned from thyroiditis to hypothyroidism and required hormone replacement therapy [95].
Autoimmune thyroiditis and Graves ophthalmopathy have been described in some case reports. For example, three patients treated with ipilimumab (two of whom in combination with bevacizumab) had autoimmune thyroiditis [96]. A 51-year-old female on ipilimumab monotherapy initially presented with periorbital swelling and pain. Initial thyroid laboratory evaluation showed normal levels of TSH and free T4 with increased anti-thyroperoxidase antibodies (anti-TPO; 662 IU/ml) and anti-thyroglobulin antibodies (148.5 IU/ml). Physical exam of the thyroid was normal, but a CT scan confirmed Graves ophthalmopathy with swelling of extraocular muscles [96]. In a patient treated with ipilimumab–bevacizumab combination therapy who presented with hand tremor, autoimmune thyroiditis was diagnosed with initial hyperthyroidism and positive anti-TPO and anti-thyroglobulin antibodies [96]. A third female patient was treated with ipilimumab and bevacizumab, and developed painless thyroiditis during treatment. She initially presented with tachycardia without goitre or neck tenderness, low TSH level (0.06 mIU/ml) and high normal free T4 level. In another report of Graves ophthalmopathy, a 51-year-old woman presented with extraocular muscle swelling and pain after ipilimumab treatment, MRI indicated potential Graves ophthalmopathy, and the levels of TSH-stimulating receptor, anti-microsomal and anti-thyroglobulin antibodies were elevated; levels of TSH and free T4 were normal [97]. Clinicians should be able to differentiate the aetiology of the thyroid dysfunction for proper management of the IRAE.
Mechanism
In some reports, polymorphisms in CTLA4 have been shown to lead to a higher incidence of autoimmune disorders, such as Graves disease and Hashimoto thyroiditis [98,99]. For example, in one study, 75% of patients with GG alleles at a single nucleotide polymorphism (SNP), JO33, developed adverse autoimmune effects, such as juvenile onset diabetes mellitus, whereas 33% of those with AA or AG alleles presented with similar adverse effects88. However, this finding was not supported in other studies looking at the link between SNPs in the CTLA4 gene and risk of primary thyroid disorder [100,101]. Although in a meta-analysis of studies evaluating 49A>G SNP this polymorphism was associated with increased susceptibility to Graves ophthalmopathy in the general Chinese population, none of the polymorphisms evaluated within individual patients was confirmed to be associated with Graves ophthalmopathy [101].
Monitoring and management
No definitive recommendations regarding monitoring for primary hypothyroidism in patients undergoing immunotherapy have been reported. In our recommendation, monitoring patients for signs of hypothyroidism, such as fatigue, weight gain and cold intolerance, is important. In addition to baseline TFTs (such as serum TSH and free T4), before initial immunotherapy, subsequent TFTs should be measured during treatment. When a patient notes any signs of thyroid dysfunction, TFTs should be measured. If evidence of hyperthyroidism or hypothyroidism is recorded, thyroid autoantibodies can also be measured [102]. Primary hypothyroidism should be treated using levothyroxine hormone replacement therapy, whereas subclinical cases would favour further observation (such as in patients who are asymptomatic with levels of TSH <10 and normal levels of free T4) [103]. In severe thyrotoxicosis, before progression to hypothyroidism, administering corticosteroid in those patients with severe symptoms might be prudent, while β-blockers could be useful for the treatment of symptoms and signs of thyrotoxicosis, such as hand tremor and tachycardia [103]. Those patients with mild symptoms of hyperthyroidism from thyroiditis can be observed and monitored for symptom progression, as well as for the development of permanent hypothyroidism.
Radioactive iodine uptake can be used to distinguish Graves disease from thyroiditis [104]. Specifically, increased uptake of iodine is consistent with Graves disease, whereas low uptake would be consistent with thyroiditis [104]. However, given the frequent use of iodinated CT contrast in those patients who undergo cancer therapy, radioactive iodine uptake might not be a very sensitive test, as the thyroid would be saturated with iodine resulting in low uptake regardless of the aetiology of hyperthyroidism [40]. In addition, a high level of serum TSH-receptor antibody and the presence of ophthalmopathy would be consistent with Graves disease instead of thyroiditis [103].
Adrenalitis
Although extremely rare, adrenalitis and subsequent primary adrenal insufficiency associated with ipilimumab therapy have been reported [105,106]. Bilateral adrenal gland enlargement after ipilimumab treatment has been reported in a patient who had normal-sized adrenal glands at baseline and simultaneous hypophysitis [105]. The patient's cortisol and aldosterone concentrations soon after the diagnosis of hypophysitis did not respond to cosyntropin stimulation, which indicates primary adrenal insufficiency; the size of the adrenal glands normalized within 6 weeks [105]. In a case report of a 79-year-old patient with metastatic melanoma, symmetric, smoothly enlarged, hypermetabolic adrenal glands were observed after 3 months of ipilimumab treatment [106]. The patient had normal-sized adrenal glands at baseline, which normalized in size and metabolic activity on follow-up scans that were performed 8 months after the end of therapy. This patient also had an elevated cortisol level at 4 months, when the adrenal gland enlargement was first noted. ACTH levels were unreported, cortisol levels were normal at 8 months [106]. When adrenal enlargement is observed in patients, assessing adrenal function through the measurement of ACTH and cortisol levels, as well as a cosyntropin stimulation test, is important to rule out primary adrenal insufficiency [57].
PD1 antibodies
Pembrolizumab and nivolumab (formerly known as MK-3476 and MDX-1106, respectively) were approved by the FDA for the treatment of patients who were already treated with ipilimumab for unresectable or metastatic melanoma and disease progression [107]. Both antibodies inhibit the interaction between PD1 and its ligands, increase the immune response against cancer cells [20,108] and are efficacious against non-small-cell lung cancer, renal cell cancer, bladder cancer and Hodgkin lymphoma [109–111]. Unlike ipilimumab, in which hypophysitis is one of the more severe and frequent endocrine adverse effects, PD1 antibody therapy has not been linked to this same increased incidence of hypophysitis, which occurs in <1% of patients [12].
Primary thyroid dysfunction
Some investigators have reported that between 39.0% and 54.2% of patients treated with PD1 antibodies experience an IRAE [11]. Of those patients, 4.7–6.0% had grade 3 or 4 endocrine adverse effects, based on the Common Terminology Criteria for Adverse Events [112]. In our literature review, the most commonly reported endocrine adverse effect with PD1 antibody therapy was hypothyroidism, with an incidence of 160 of 2,573 cases [11,12,26,66,67,107,109,113–117] (∼5.9%; Table 3). Hyperthyroidism was recorded in 1.0–4.7% of patients (mean 3.3%; 71 of 2,153 cases). As specific clinical presentations, laboratory test results and subsequent management of thyroid dysfunction in these patients were not discussed in these clinical trials, we are unable to determine the precise aetiologies of these. The time to occurrence of thyroid abnormalities was not indicated in the trials that we reviewed; however, primary hypothyroidism has been reported to present between 5 months to 3 years after PD1 antibody treatment, but these data include a ipilimumab–nivolumab combination trial [32]. In a report of 10 patients with nivolumab-related thyroiditis, individuals had abnormal TFTs ∼3–8 weeks after the first dose of nivolumab [118]. Six of these 10 cases of thyroiditis were observed during the transient thyrotoxic phase. TSH receptor antibodies were absent in all patients, but four individuals were positive for thyrotropin-binding inhibitory immunoglobulins and/or TPO antibodies [118]. Along with low levels of TSH and elevated free T4 and T3, clinical presentations included fatigue and palpitations 3–6 weeks after the first anti-PD1 treatment. These patients all developed hypothyroidism, which required thyroid hormone replacement therapy. In the remaining 4 of 10 cases, hypothyroidism was found without the thyrotoxic phase ∼6–8 weeks after initial nivolumab treatment; of these, three patients also had anti-thyroglobulin and two had anti-TPO antibodies present in their serum. This finding suggests that, although both hypothyroidism groups have a common disease process, those with hypothyroidism but without thyrotoxicosis have had a subclinical presentation of the thyrotoxic phase, which is therefore undetected.
In another report, two possible cases of subclinical autoimmune thyroid dysfunction in patients undergoing nivolumab therapy were detailed [119]. Both cases reported thyroid ultrasonography findings that were consistent with Hashimoto thyroiditis, with elevated levels of anti-thyroglobulin and TPO antibodies. One of these patients developed worsening primary hypothyroidism after the second administration of nivolumab, whereas the other patient presented with initial hyperthyroidism immediately before the second administration of this anti-PD1 therapy. Unfortunately, this latter patient had limited follow-up, and therefore this patient's ultimate thyroid status is unknown [119].
Painless thyroiditis has also been reported in a 55-year-old woman who developed dyspnoea, anxiety diarrhoea and palpitations 3 weeks after her second nivolumab treatment; the levels of TSH were undetectable, and the levels of free T4 (2.06 ng/dl) and T3 (554.2 pg/dl) were high [120]. Thyroid autoantibody assays showed that the patient had normal levels of TPO antibodies and normal thyroid-stimulation immunoglobulins with elevated thyroglobulin antibodies, which indicated a level of thyroid autoimmunity; ultrasonography, which usually is not a diagnostic test for thyroiditis and can appear normal in mild cases, showed normal vasculature and density within the thyroid gland [120]. This patient returned to a euthyroid state after therapy with β-blockers and withdrawal of immunotherapy, but it is important to note that thyrotoxicosis can still progress to a hypothyroid state, as described above [118]. Given this finding, nivolumab-induced hypothyroidism is likely to be sequela of thyroiditis.
Mechanism
The mechanism that is responsible for nivolumab-induced thyroid dysfunction is unclear. In a case series of patients presenting with painless thyroiditis and hypothyroidism after PD1 antibody therapy, 8 of 10 patients were positive for anti-thyroglobulin and anti-thyroid peroxidase antibodies, and all were negative for thyrotropin-binding inhibitory immunoglobulins [118]. Although not verified, the investigators hypothesized that polymorphic variants in the PD1 gene in some individuals might predispose them to an increased risk of endocrine dysfunction [118]. Future studies to understand the mechanism by which PD1 antibodies affect the thyroid tissue are necessary.
Management
When faced with transient thyrotoxicosis, physicians should act rapidly to ensure the best outcomes for their patients. Supportive therapy with β-blockers can help to alleviate adrenergic symptoms and signs of hyperthyroidism, and immunotherapy could be held if severe symptoms are present [120]. Radioactive iodine uptake can be inaccurate given the frequency with which iodine contrast-enhanced imaging is used in patients with cancer [40]. In the thyrotoxic phase of thyroiditis, TSH is expected to be low, and the levels of free T4 or T3 are expected to be elevated [85]. Immunoassays to detect levels of TPO and thyroglobulin antibodies can be used to understand the autoimmune aetiology of thyroiditis [118]. Certain TSH receptor antibodies, such as thyrotropin-stimulating immunoglobulins, can be used to distinguish Hashimoto thyroiditis from Graves disease to direct appropriate management [121].
If thyroiditis is in the thyrotoxic phase, the patient should be monitored for symptoms, signs and laboratory test abnormalities consistent with progression to hypothyroidism. Levothyroxine hormone replacement should be administered to treat overt hypothyroidism [12]. Patients who have yet to experience this IRAE should be monitored for any thyroid dysfunction. Although no established schedule for monitoring for thyroid dysfunction in those individuals undergoing anti-PD1 therapy exists, a close observation for signs of hyperthyroidism or hypothyroidism during follow-up visits is recommended. We further recommend that TFTs are monitored in those patients undergoing anti-PD1 therapy. Given that primary thyroid dysfunction can present as early as 3 weeks and as late as 3 years after treatment [32,120], TFTs should be monitored monthly for the first 6 months of treatment. For the management of CTLA4 immunotherapy, we recommend that, for those individuals receiving PD1 antibodies, if thyroid function is normal and the patient is asymptomatic, testing could be done every 3 months for 6–12 months followed by further assessment every 6 months for years 2 and 3. When patients have symptoms or signs of thyroid dysfunction, they should have a prompt evaluation for anti-PD1-induced thyroid dysfunction, especially given that fatigue is the most common overall adverse effect in patients receiving PD1 blockade.
Hyperglycaemia
Although rare, PD1 pathway blockade can lead to diabetes mellitus [122–125]. Anti-PD1 therapy was responsible for eight cases of type 1 diabetes mellitus (T1DM), with an additional case that was observed in a patient who was treated with anti-PDL1 therapy. Seven of nine patients with T1DM initially presented with diabetic ketoacidosis (DKA), with the remaining two patients presenting with severe hyperglycaemia [122–125]. Overall, the presence of glutamic acid decarboxylase 65 (GAD65) antibodies, a marker of T1DM along with DKA, was found in five patients undergoing nivolumab treatment, whereas three patients either had DKA or were GAD65 positive [122–125].
Although the mechanism underlying the onset of T1DM in patients receiving PD1 immunotherapy is not well understood, the modulation of T cell regulatory function has been suggested to be responsible for this IRAE124. Specifically, three of five patients were found to have T1DM-specific autoantibodies (GAD65), and two of five patients presented with upregulation of CD8+ T cell response to a T1DM antigen [124].
Although patients with PD1 immunotherapy-related T1DM should be treated with insulin, no screening guidelines have been established to detect T1DM in these patients. Given the morbidity related to DKA and hyperglycaemia, clinicians managing patients undergoing anti-PD1 therapy should carefully monitor patients for elevated levels of blood sugar.
PDL1 antibodies
Endocrine-related IRAEs in response to PDL1 antibodies should also be managed using hormone replacement therapies, and reintroduction to treatment, when appropriate26. Mechanistically, polymorphisms in genes encoding PDL1 or PD1 might increase the susceptibility to autoimmune disease [126–128]. In a phase I clinical trial of the PDL1 antibody MDX-1105, adverse effects occurred in 39% of the 207 patient cohort, with only a small number (n = 10; 5%) presenting with incidences above grade 3 (Ref. 26). Hypothyroidism was observed in six of 207 patients (3%), adrenal insufficiency in two patients (1%) and autoimmune thyroiditis in another 1% of the cohort [26]. In these studies, clinical presentation or laboratory test results were not reported to determine if hypothyroidism was secondary to thyroiditis, if adrenal insufficiency was primary or secondary or to specify the nature of the autoimmune thyroiditis. Phase I trials of atezolizumab and durvalumab also had elevated incidences of IRAEs compared with trials of MDX-1105. Specifically, hypothyroidism was reported in six of 70 patients in the atezolizumab cohort (9%) and 10 of 99 in the durvalumab cohort (10.1%) [117,129]. Although the elevated incidence of hypothyroidism with durvalumab might be partially explained by the combined administration with tremelimumab in this trial, the reason why atezolizumab would have an increased incidence of hypothyroidism is unclear. Further, specific clinical and biochemical presentations were not reported, which makes determining the aetiology of the hypothyroidism difficult.
When comparing the adverse effects of PDL1 and PD1 antibody therapy, the overall adverse effect rates (41% in PD1; 30% in PDL1) and endocrine-related rates (4% for both PD1 and PDL1) were similar12,26. However, hypothyroidism was reported in 4.3% of the PDL1 cohort and in 5.9% of the PD1 cohort [11,12,26]. Although reports of thyroiditis were rare (∼1%) in patients from the PDL1 cohort and limited to individual case reports in the PD1 cohort [11,12,26,119,120], this finding could be attributable to patients being misclassified as having hypothyroidism and not as thyroiditis. The incidence of thyroiditis might therefore be higher than that reported in the literature.
PDL1 also binds to CD80 in addition to its interaction with PD1 receptors on activated T cells, which might explain the differences in the reactions to PDL1 and PD1 inhibitors. Specifically, PDL1 antibodies would affect the interaction of its target ligand with both CD80 and PD1 receptors, whereas blockade of PD1 would result in inhibition of its interaction with PDL1 and PDL2 (a subtype of B7 family ligands that is related to PDL1) ligands [130]. This complex set of ligand–receptor interactions by PD1 and PDL1 might account for the differences in incidence of endocrine-related adverse effects [130].
Combination therapies
Combined CTLA4 and PD1 blockade has been explored in preclinical models and clinical trials in metastatic melanoma [5,50,131–133]. In our review of combination therapy clinical trials, the incidence of hypothyroidism was 13.9% (64 of 462 cases) and that of hypophysitis was 8.0% (37 of 462 cases; Table 4) in patients who received this double therapy. In a phase I trial, the combination of ipilimumab and nivolumab had incidences of 13% primary hypothyroidism (6 of 45 patients), 9% thyroiditis (4 of 45) and 9% hypophysitis (4 of 45) [32]. These data suggest that additive effects of the two therapies contribute to certain IRAEs. Moreover, 53% of the cohort undergoing concurrent administration of nivolumab and ipilimumab presented with grades 3 to 4 incidences of all treatment-related adverse effects, compared with only 18% of the cohort who received the treatment sequentially, 20% in the ipilimumab-only study and 15% in the nivolumab-only study [7,12,50]. In a phase II trial of concurrent administration of nivolumab and ipilimumab, a significantly higher rate of objective response (61%) in the combination group than in the ipilimumab monotherapy group (11%) was reported (P< 0.001) [131]. As seen in previous combination therapy trials, the incidence of grade 3 to 4 adverse effects was elevated at 54% in the combination cohort versus 24% in the monotherapy group. Specifically, hypophysitis was observed in 12% (7% in monotherapy group) and hypothyroidism in 16% (15% in monotherapy group). Similarly, in a subgroup analysis of CheckMate 067 phase III trial evaluating the efficacy of nivolumab and ipilimumab, an increased incidence of overall grade 3 to 4 adverse effects in the combination group (55%) was seen compared with 16.3% in the nivolumab-only group and 27.3% in the ipilimumab-only group [134]. In those patients who received more than one immunotherapeutic agent, we recommend that the clinician assess thyroid, pituitary and adrenal function, as well as the presence of hyperglycaemia, as described for monotherapy earlier in the text. The additive adverse effects of combination checkpoint blockade therapy should be further examined.
Adverse effects and treatment response
The clinical response to immunotherapy has been associated with the occurrence of IRAEs [38,64]. In one study, 3 of the 14 patients in the cohort who had a clinical response to CTLA4–IL-2 combination therapy also had grade 3–4 IRAE toxicities [64]. Furthermore, other investigators describe two adjuvant ipilimumab trials for patients with stage 3–4 melanoma in which a significant correlation exists between relapse-free survival and presentation of IRAEs [38,135]. Finally, patients presenting with ipilimumab-related hypophysitis have been reported to have a median survival time of 19.4 months compared with a median survival time of 8.8 months for those not presenting with hypophysitis (P = 0.05) [36]. Although encouraging, the investigators postulate that this analysis might be biased, as those who do not survive long enough during the trial do not get the long-term observation of patients with IRAEs [36]. Further analysis of the relationship between IRAEs and clinical outcomes must be conducted to understand benefits, if any, of such phenomena.
Conclusions
Major advances in the understanding of the immune response in cancer have led to rapid progress in clinical immunotherapy trials in the past decade. Although immunotherapies lack the traditional profile of chemotherapy- related adverse effects, a rare, but major, set of IRAEs has emerged. In clinical trials, an increased susceptibility to hypophysitis in those treated with CTLA4-targeted immunotherapy has been revealed. PD1-targeted treatments have been predominately linked with primary thyroid dysfunction, with a few rare cases of T1DM. Despite the current clinical understanding of endocrine IRAEs, which has led to effective treatment strategies with hormonal replacement, additional studies are needed to further understand the clinical characteristics and chronology of these adverse effects and to clarify the mechanism by which immunotherapy results in endocrinopathies.