ICPis-Induced Autoimmune Polyendocrine Syndrome Type 2: A Review of the Literature and a Protocol for Optimal Management

Abstract

Context

Immune checkpoint inhibitors (ICPis) targeting cytotoxic T-lymphocyte antigen 4 (CTLA-4), programmed cell death protein 1 (PD-1), and its ligand (PD-L1) are now approved to treat a variety of cancers. However, ICPis therapy is associated with a risk of immune-related adverse events (irAEs). Autoimmune polyendocrine syndrome type 2 (APS-2) is a rare endocrine irAE.

Evidence Acquisition

Several databases (PubMed, Web of Science, Cochrane Central Registry of Controlled Trials, ClinicalTrials.gov, and Scopus) were searched up to February 18, 2020, for case reports on endocrine irAEs and ICPis. The reported side effects and adverse events of the ICPis therapy in the US Food and Drug Administration (FDA) and European Medicines Agency (EMA) adverse events pharmacovigilance registries are also included.

Evidence Synthesis

Here, we provide an overview of all published and reported cases (n = 30) of ICPis-induced APS-2. We summarize the clinical characteristics, autoantibodies, human leukocyte antigen (HLA) genotypes, and therapies and propose an APS-2 screening strategy.

Conclusions

Given the life-threatening risks of endocrine dysfunction if it is not promptly recognized (such as diabetic ketoacidosis and acute adrenal crisis), physicians (especially endocrinologists and oncologists) should be familiar with APS-2. After diagnosis of an autoimmune disease induced by ICPis (especially PD-1 inhibitors), patients with a high-risk HLA allele (HLA-DR4) require close monitoring for the development of APS-2.

immune checkpoint inhibitors, immune-related adverse events, autoimmune polyendocrine syndrome

Immune checkpoint inhibitors (ICPis) targeting cytotoxic T-lymphocyte antigen 4 (CTLA-4), programmed cell death protein 1 (PD-1), and its ligand (PD-L1) are now approved to treat a variety of cancers, such as melanoma and lung, kidney, and liver tumors (1). Physiologically, CTLA-4, PD-1, and PD-L1 play key roles in peripheral tolerance. Thus, pharmacologic disruption of these checkpoints can trigger autoimmune-like manifestations in various organ systems, generally referred to as immune-related adverse events (irAEs) (2, 3). Endocrine disorders (hypothyroidism, hyperthyroidism, thyroiditis, hypophysitis, primary adrenal insufficiency, and type 1 diabetes [T1DM]) are common irAEs that have been reported in clinical trials of ICPis (4, 5). However, previous studies mainly focused on the incidence of single endocrine irAEs. Individuals with one autoimmune disorder are at higher risk of a second autoimmune disorder (6, 7). The growth in the number of active clinical trials testing anti-PD-1/PD-L1 agents has been dramatic, increasing from a single trial in 2006 to 2250 trials as of September 2018, and including more than 380 900 participants (8). Thus, given the increasing use of ICPis therapy in clinical practice, cases of ICPis-induced polyendocrinopathies will increasingly be reported.

Autoimmune polyendocrine syndrome (APS) comprise a diverse group of clinical conditions characterized by functional impairment of multiple endocrine tissues due to the loss of immune tolerance (9). APS can be categorized as rare monogenic forms, such as APS type 1 (APS-1), or a more common polygenic variety known as APS type 2 (APS-2). Patients with APS-2 have disease courses characterized by at least 2 of the following 3 endocrinopathies: T1DM, autoimmune thyroid disease (AITD), and Addison’s disease (9).

Given the life-threatening risks of endocrine dysfunction if it is not promptly recognized (such as diabetic ketoacidosis [DKA] and acute adrenal crisis), here, we provide an overview of all published and reported cases (n = 30) of ICPis-induced APS-2.

Materials and Methods

Several databases (PubMed, Web of Science, Cochrane Central Registry of Controlled Trials, ClinicalTrials.gov, and Scopus) were searched up to February 18, 2020, for case reports on ICPis and irAEs in patients with advanced solid tumors. The reported side effects and adverse events of the ICPis therapy in the US Food and Drug Administration (FDA) and European Medicines Agency (EMA) adverse events pharmacovigilance registries are also included. Studies written in languages other than English were not considered for inclusion in this review. Studies listed in the reference sections of the studies retrieved during the literature search were reviewed in order to identify additional studies. The search strategy involved the following terms: “Nivolumab” or “Pembrolizumab” or “Cemiplimab” or “Toripalimab” or “Sintilimab” or “Atezolizumab” or “Avelumab” or “Durvalumab” or “Ipilimumab” or “Tremelimumab” or “Programmed cell death 1 receptor” or “Cytotoxic T-lymphocyte antigen 4” or “Programmed cell death 1 receptor ligand” or “PD-1”or “PD-L1” or “CTLA-4” and “Endocrinopathy” or “Autoimmune polyendocrine syndrome” or “Autoimmune polyglandular syndrome” or “Schmidt syndrome” or “Polyendocrinopathies” or “Autoimmune” or “Endocrinopathies” or“T1DM”or“Diabetes Mellitus” or “Type 1 diabetes” or “Autoimmune diabetes ”or “Hyperthyroidism” or “Hypothyroidism” or “Thyroiditis” or “Thyroid disease” or “Addison’s disease” or “Primary adrenal insufficiency.” The following data were extracted from each manuscript: first author, year of publication, age, gender, country and ethnicity of the patient, cancer type, ICPis therapy, number of cycles of therapy and duration, prior immunotherapy, relevant medical history, presence of DKA, plasma glucose level at first diagnosis, glycated hemoglobin, C-peptide, autoantibodies, irAEs, treatment of irAEs, tumor response, and human leukocyte antigen (HLA) genotype. The consort flowchart of the literature search, identification, and selection is shown in Fig. 1.

Figure 1.

The CONSORT flowchart of the literature search, identification and selection.

Results

Among all published and reported cases (n = 30) of ICPis-induced APS-2, 26 cases of ICPis-induced APS-2 have been reported in the literature. Tables 1 and 2 summarize the key findings. In addition, 5 cases were reported in the FDA and EMA adverse events pharmacovigilance registries (Table 3), 1 of which was the patient reported by Jared Lowe et al (19) (case 10 in Table 1).

Table 1.

Reported cases

Patient	First author, year [reference]	Tumor type	Immune checkpoint inhibitors (ICPis)	Age (years)	Sex	Country
1	Giulia Lanzolla, 2019 (10)	Lung adenocarcinoma	Atezolizumab	60	Male	Italy
2	Ashray Gunjur, 2019 (11)	Melanoma	Pembrolizumab	78	Female	Italy
3	Shivani Patel, 2019 (12)	Lung adenocarcinoma	Durvalumab	49	Female	Australia
4	Kanako Sakurai, 2018 (13)	RCC	Nivolumab	68	Female	Japan
5	Li Li, 2017 (14)	SCC	Nivolumab	63	Male	USA (African American)
6	Elizabeth Hansen, 2016 (15)	Melanoma	Pembrolizumab	58	Male	USA
7	Mahnaz Mellati, 2015 (16)	Sarcomatoid SCC (jaw)	PD-1 inhibitor	66	Female	USA
8	Marie-Léa Gauci, 2017 (17)	Melanoma	Nivolumab	73	Male	France
9	Lars Hofmann, 2016 (18)	Melanoma	Nivolumab	70	Female	Germany
10	Jared Lowe, 2016 (19)	Melanoma	Ipilimumab+Nivolumab	54	Male	USA
11	Jing Hughes, 2015 (20)	Melanoma	Ipilimumab+Nivolumab	55	Female	USA
12	Jing Hughes, 2015 (20)	Melanoma	Pembrolizumab	64	Female	USA
13	Caroline Gaudy, 2015 (21)	Melanoma	Pembrolizumab	44	Female	France
14	Sung Hye Kong, 2016 (22)	SCC	Pembrolizumab	68	Male	Korea
15	M. Alhusseini, 2017 (23)	Lung adenocarcinoma	Ipilimumab+pembrolizumab	65	Male	USA
16	Anne-Cécile Paepegaey, 2017 (24)	Melanoma	Pembrolizumab	55	Female	France
17	Emma Scott, 2018 (25)	Melanoma	Ipilimumab+pembrolizumab	58	Male	Australia
18	Sumumu Kurihara, 2020 (26)	Parotid gland adenocarcinoma	Nivolumab	48	Male	Japan
19	Jeroen de Filette, 2019 (27)	NSCLC	Pembrolizumab	61	Male	Belgium
20	A Ram Hong, 2020 (28)	Lung cancer	Pembrolizumab	76	Male	Korea
21	Osamah Hakami, 2019 (29)	Melanoma	Pembrolizumab	52	Male	Ireland
22	A Galligan, 2018 (30)	SCC (oropharynx)	Pembrolizumab	82	Male	Australia
23	A Galligan, 2018 (30)	Melanoma	Pembrolizumab	23	Male	Australia
24	Lucien Marchand, 2019 (31)	Melanoma	Pembrolizumab	83	Male	Belgium
25	Lucien Marchand, 2019 (31)	Melanoma	Pembrolizumab	65	Male	Belgium
26	S Hescot, 2018 (32)	SCC (cervical)	Pembrolizumab	33	Female	France
Patient	Conditions	Autoantibody found	HLA type		Treatment withdrawal	Response
1	T1DM, AD, hypophysitis	Anti-21-hydroxylase	DRB1*04, DQB1*03		Yes	PD
2	T1DM, AD, thyroid disease	GADA, IA-2A	DRB1*04.16, DQB1*02.05, DQA1*01.03		No	CR
3	T1DM, thyroid disease	GADA	NR		No	CR
4	T1DM, thyroid disease	TPOAb, TgAb	DRB1*09:01, DQB1*03:03		No	CR
5	T1DM, thyroid disease	GADA, TPOAb	NR		Yes	PD
6	T1DM, thyroid disease	GADA	NR		Yes	CR
7	T1DM, thyroid disease	GADA	DR3-DQ2, DR4-DQ8		No	NR
8	T1DM, thyroid diseasea	GADA, ZnT8A	NR		Yes	CR
9	T1DM, thyroid disease	-	NR		No	CR
10	T1DM, AD, thyroid disease, hypophysitis	GADA, TRAb	HLA-I A2, HLA-II DQB1*0602		No	CR
11	T1DM, thyroid diseasea	-	A2.1+, DR4+		No	NR
12	T1DM, thyroid diseasea	-	DR4+		No	NR
13	T1DM, thyroid diseasea	-	NR		No	CR
14	T1DM, thyroid disease	-	DRB1*09:01, DQB1*03:03		No	PR
15	T1DM, thyroid disease	GADA, ICA, IA-2A, TPOAb	NR		b	PR
16	AD, thyroid disease	anti-21-hydroxylase, adrenal cortex antibodies	NR		No	PD
17	T1DM, thyroid disease	-	NR		No	NR
18	T1DM, thyroid disease	TRAb	DRB1*04:05		Yes	PD
19	T1DM, thyroid disease	GADA	DRB1*04, DQA1*03:01, DQB1*03:02		NR	NR
20	T1DM, AD	-	NR		Yes	PR
21	T1DM, thyroid disease	-	NR		No	CR
22	T1DM, thyroid disease	-	DRB1*04 (DR4), DQB1*03:02 (DQ8)		No	PR
23	T1DM, thyroid disease	GADA, TPOAb	DRB1*03 (DR3), DRB1*04 (DR4), DQB1*03:02 (DQ8)		Yes	PR
24	T1DM, thyroid disease	TPOAb	DRB1*01:01 DQA1*01 DQB1*05:01/DRB1*16:01 DQA1*01 DQB1*05:02”		No	PR
25	T1DM, thyroid disease	TPOAb	DRB1*04:01 DQA1*02 DQB1*02:02/DRB1*07:01 DQA1*03 DQB1*03:01		Yes	PD
26	AD, thyroid disease	TPOAb	NR		No	PR

Abbreviations: AD, Addison’s disease; CR, complete response; GADA, glutamic acid decarboxylase antibody; HLA, human leukocyte antigen; IA-2A, insulinoma-associated protein 2 antibody; NR, not reported; NSCLC, non-small-cell lung carcinoma; PD, progressive disease; PR, partial response; PD-1, programmed cell death-1; PD-L1, programmed cell death-ligand-1; RCC, renal cell carcinoma; SCC, squamous cell carcinoma; T1DM, type 1 diabetes mellitus; TPOAb, thyroid peroxidase antibody, TgAb, thyroglobulin antibody; TRAb, thyroid-stimulating hormone receptor antibody; ZnT8A, zinc transporter 8 antibody.

^aAutoimmune condition preceded treatment with immune checkpoint inhibitors (ICPis).

^bIpilimumab withdrawn but pembrolizumab continued.

Table 2.

Characteristics of 26 Patients With ICPis-Induced APS-2

Characteristic	Value
Causative agent
Anti-PD-1	20
Pembrolizumab	14
Nivolumab	5
Not clear	1
Anti-PD-L1	2
Anti-CTLA-4+anti-PD-1	4
Conditions
T1DM+AD	2
T1DM+thyroid disease	20
T1DM+AD+thyroid disease	2
AD+thyroid disease	2
Demographics
Sex (men/women)	16/10
Race (Asian/Caucasian)	4/22
Age, years	62 (23, 83)
Tumor type
Lung adenocarcinoma	3
Melanoma	14
Lung cancer	1
NSCLC	1
Parotid gland adenocarcinoma	1
RCC	1
Sarcomatoid SCC (jaw)	1
SCC	2
SCC (oropharynx)	1
SCC (cervical)	1
Interval to onset after ICPis therapy initiation
Number of cycles of therapy	4 (1-17)
Duration to onset, weeks	11.5 (3-52)
Autoantibodies
β-cell antibodies, n positive/n tested (%)
GADA	10/24 (41.7%)
ZnT8A	1/9 (11.1%)
ICA	1/10 (10%)
IA-2A	3/19 (15.8%)
IAA	1/9 (11.1%)
≥1 antibodies	10/24 (41.7%)
≥2 antibodies	3/24 (12.5%)
Thyroid autoantibodies, n positive/n tested (%)
TPOAb	6/26 (23.1%)
TgAb	1/26 (3.8%)
TRAb	2/26 (7.7%)
≥1 antibody	8/26 (30.8%)
Anti-21-hydroxylase	2
Adrenal cortex antibody	1
Response (PD/PR/CR/NR)	5/7/9/5
History of autoimmune disease	4
History of T2DM	1
Treatment withdrawal (yes/no)	9/16

Data are shown as median (range) or frequency (%) unless otherwise noted.

Abbreviations, AD, Addison’s disease; APS-2, autoimmune polyendocrine syndrome type 2; CR, complete response; CTLA-4, cytotoxic T-lymphocyte antigen 4; IAA, insulin antibody; IA-2A, insulinoma-associated protein 2 antibody; GADA, glutamic acid decarboxylase antibody; NR, not reported; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus; TPOAb, thyroid peroxidase antibody, TgAb, thyroglobulin antibody; PD, progressive disease; PR, partial response; PD-1, programmed cell death-1; PD-L1, programmed cell death ligand-1; TRAb, thyroid-stimulating hormone receptor antibody; ZnT8A, zinc transporter 8 antibody.

Table 3.

APS-2 Reported in FDA and EMA Adverse Events Pharmacovigilance Registries

Case ID	Pharmacovigilance registries, year	Tumor type	Immune checkpoint inhibitor (ICPi)	Age (years)	Sex	Country	Conditions
17336256	FDA, 2020	Malignant Melanoma	Nivolumab; Ipilimumab	80	F	Switzerland	Thyroid disease, T1DM
13160288a	FDA, 2017	Malignant Melanoma	Nivolumab; Ipilimumab	54	M	USA	Thyroid disease, Adrenal Insufficiency, Hypophysitis, Fulminant T1DM
EU-EC- 10002225803	EMA, 2018	Intraocular melanoma	Ipilimumab; Pembrolizumab	NR	F	NR	Thyroid disease, T1DM
EU-EC- 10004524922	EMA, 2019	Renal cancer	Nivolumab; Ipilimumab	49	M	NR	AD, thyroid disease
EU-EC- 10004247855	EMA, 2019	Metastatic malignant melanoma	Nivolumab; Ipilimumab	68	M	USA	AD, thyroid disease

Abbreviations: AD, Addison’s disease; NR, not reported; T1DM, type 1 diabetes mellitus.

^aThe patient was reported by Jared Lowe (case 10 in Table 1).

Clinical characteristics of patients with ICPis-induced APS-2

Among the 26 cases, there were 16 men and 10 women. They had a median age at APS-2 diagnosis of 62 years (Table 2). One of the patients had a history of type 2 diabetes mellitus. There was a history of autoimmune thyroid diseases in 15.4% (4/26) of the patients.

The patients were diagnosed with APS-2 after a median of 4 treatment cycles (range, 1-17) and after a median duration of 11.5 weeks (range, 3-52). The treatment regimens comprised PD-1 inhibitor monotherapy (20/26, 76.9%) (pembrolizumab [70.0%, 14/20], nivolumab [25.0%, 5/20], and not clear [5%, 1/20]); PD-L1 inhibitor monotherapy (2/26, 7.7%); and a combination of CTLA-4 inhibitors and PD-1 inhibitors (4/26, 15.4%). The most common treatment regimen was PD-1 inhibitor monotherapy; blockade of the PD-1/PD-L1 pathway was involved in 100% of the cases.

The majority (24/26) of the cases involved primary thyroid disease that presented as hypothyroidism, hyperthyroidism, or thyroiditis. In addition, 92.3% (24/26) had T1DM and only 23.1% (6/26) had Addison’s disease. Two cases had all 3 of the endocrinopathies involved in APS-2, while 92.3% (24/26) of the cases had 2 (mainly thyroid disease and T1DM [20/26]).

T1DM diagnosis during treatment was based on the presence of hyperglycemia, DKA, low or undetected serum C-peptide at diagnosis or during follow-up, and positive beta-cell antibodies. Among 24 cases of T1DM, 41.7% (10/24) cases were positive for at least one β-cell antibody, 71.4% (15/21) cases presented with DKA, and 85% (17/20) cases had low or undetected serum C-peptide (<100 pmol/L) at diagnosis or during the follow-up. Furthermore, 91.7% (22/24) cases presented with DKA, or low or undetected serum C-peptide (<100 pmol/L) at diagnosis or during the follow-up, or were positive for at least one β-cell antibody (Fig. 2). The other 2 cases were as follows: Case 12 presented with ketonuria, glucose of 39.06 mmol/L, glycated hemoglobin A1c (HbA1c) of 7.4%, 166.5 pmol/L (0.5 ng/mL) of random C-peptide, and HLA-DR4; Case 24 had no symptoms, glucose of 33 mmol/L, HbA1c of 9.4%, and a serum C-peptide level of 333 pmol/L. The patient remained insulin-dependent at last follow-up. T1DM diagnosis during treatment was mainly based on the presence of hyperglycemia, DKA, low or undetected serum C-peptide at diagnosis or during follow-up, and positive β-cell antibodies.

Figure 2.

Diagnosis of T1DM during treatment: among 24 cases of T1DM, 10 cases were positive for at least one beta-cell antibody, 15 cases presented with DKA, and 17 cases had low or undetected serum C-peptide (<100 pmol/L) at diagnosis or during the follow-up. 6 cases presented with both positive beta-cell antibody and low or undetected C-peptide; 7 cases were diagnosed based on positive beta-cell antibody and DKA; 11 cases presented as DKA and low or undetected C-peptide; 4 cases presented with positive beta-cell antibody, low or undetected C-peptide and DKA. Abbreviations: Ab+, positive beta-cell antibody; DKA, diabetic ketoacidosis.

In addition, with the exception of the duplicated case, there were 2 cases reported in each of the FDA (n = 96022) and EMA (n = 6861) adverse events pharmacovigilance registries. These 4 cases included 2 men and 2 women, who received a combination of PD-1 inhibitor and CTLA-4 inhibitor. All 4 of these patients had thyroid diseases, 2 had T1DM and 2 had Addison’s disease (Table 3).

Autoantibodies and HLA genotype

Among the 26 cases, 30.8% (8/26) were positive for at least one thyroid antibody. Of the 24 T1DM patients, 41.7% (10/24) had at least one β-cell autoantibody, and 12.5% (3/24) had at least two β-cell autoantibodies. 41.7% (10/24) of the patients had glutamic acid decarboxylase antibodies (GADA), 15.8% (3/19) of the patients had insulinoma-associated protein-2 antibodies (IA-2A), 11.1% (1/9) of the patients had zinc transporter 8 antibodies (ZnT8A), 11.1% (1/9) of the patients had insulin antibodies (IAA), 10% (1/10) of the patients had islet cell antibodies (ICA), 2 patients had anti-21-hydroxylase (2/2,100%), and 1 patient had adrenal cortex antibody (1/1, 100%).

Of the 26 cases, 14 were HLA genotyped; among these individuals, 71.4% (10/14) were HLA-DR4.

Treatment of ICPis-induced APS-2

The treatment for immunotherapy-induced diabetes and DKA is standard insulin therapy. After starting insulin therapy, glycemic control was reached in almost all cases (n = 22). However, 2 cases involved challenging control, with severe instability of blood glucose levels. The immunotherapy did not cause a change in plasma glucose levels, and the insulin requirements and fasting blood glucose levels were maintained. Only 1 patient who developed hyperthyroidism was treated with methimazole, while the others required no treatment. Almost all of the patients with hypothyroidism required ongoing treatment with thyroxine (n = 1) or levothyroxine (n = 9). Of the 6 cases with Addison’s disease, 5 started taking exogenous corticosteroids, with fludrocortisone replacement therapy being used in 3 of them.

Endocrinopathies involved in ICPis-induced APS-2

Among the 26 cases, 20 had thyroid disease and T1DM concurrently or successively (Table 4). Of these 20 cases, 8 presented with T1DM and 6 presented with thyroid disease as the first endocrine disorder, but this information was not reported for the remaining 6 cases. Furthermore, 66.7% (12/18) of the patients presented with DKA, with a median plasma glucose level at diagnosis of 32 mmol/L (range, 10.5-66.3) and a median glycated hemoglobin level of 8.1% (range, 6.4-9.7). Undetectable or low C-peptide levels were present at diagnosis in 86.7% (13/15) of the tested patients. The mean interval to onset was 3 cycles (range, 1-17) and 7.5 weeks (range, 3-52) for the antibody-positive patients, and the median interval to onset was 5 cycles (range, 2-12) and 13 weeks (range, 4-23) for the antibody-negative patients. Of the 20 cases of ICPis-induced thyroid disease, 55% (11/20) involved thyroiditis, 25% (5/20) involved primary hypothyroidism, and 20% (4/20) involved hyperthyroidism. A summary of the results is shown in Table 4.

Table 4.

Characteristics of 20 Patients With T1DM and Thyroid Disease

Characteristic	Value
First diagnosis (T1DM/thyroid disease/NR)	8/6/6
Interval to onset after ICPis therapy initiation
T1DM – Positive for ≥1 β-cell antibody
Number of cycles of therapy	3 (1-17)
Duration to onset, weeks	7.5 (3-52)
– Negative for β-cell antibodies
Number of doses	5 (2-12)
Duration to onset, weeks	13 (4-23)
Thyroid disease
Number of doses	2 (1-6)
Duration to onset, weeks	6 (2-13)
Glucose at diagnosis, mmol/L	32 (10.5–66.3)
HbA1c, %	8.1 (6.4–9.7)
Undetectable or low serum C-peptide (<100 pmol/L)	13/15 (86.7%)
DKA	12/18 (66.7%)
FT1DM	6/20 (30%)
Thyroid function
Hyperthyroidism	4
Hypothyroidism	5
Thyroiditis	11

Data are shown as median (range) or frequency (%) unless otherwise noted.

Abbreviations: DKA, diabetic ketoacidosis; FT1DM, fulminant diabetes mellitus; NR, not reported; T1DM, type 1 diabetes mellitus.

Antitumor response

The patients undergoing ICPis therapy mainly had metastatic lung cancer or melanoma. After APS-2 diagnosis, 61.5% (16/26) of the patients continued taking the ICPis therapy. Overall, 76.2% (16/21) of the patients who developed ICPis-induced APS-2 had a partial or complete antitumor response.

Proposal of a screening strategy for APS-2

Clinicians (especially endocrinologists and oncologists) should be aware of endocrine irAEs, including single endocrine events (disorders of the thyroid, islets, adrenal glands, or pituitary gland) and polyendocrine events. To minimize long delays in APS-2 diagnosis and to avoid the development of life-threatening DKA or adrenal crisis, we propose screening for APS-2 in patients receiving ICPis therapy, as shown in Fig. 3.

Figure 3.

Proposal of education, monitoring, screening for other ICPis-induced autoimmune diseases after the diagnosis of an initial ICPis-induced autoimmune disease, diagnosis and treatment, and follow-up related to APS-2 in patients taking ICPis therapy. *Clinical alert signs (34): headaches that will not go away or unusual headache patterns; vision changes; rapid heartbeat; increased sweating; extreme tiredness or weakness; muscle aches; weight gain or loss; dizziness or fainting; feeling more hungry or thirsty than usual; hair loss; changes in mood or behavior (such as decreased sex drive, irritability, or forgetfulness); feeling cold; constipation; deeper voice; urinating more often than usual; nausea or vomiting; and abdominal pain # Step 4: Screening for APS-2 Further testing needed (33, 34) 1)Islets: C-peptide level, β-cell antibody 2)Thyroid: thyroglobulin antibody (TgAb), thyroid peroxidase antibody (TPOAb), thyroid-stimulating hormone receptor antibody (TRAb) +/- thyroid ultrasound +/- thyroid scintigraphy 3)Adrenal glands: cosyntropin or synacthen stimulation test, +/- adrenal computed tomography scan, anti-21-hydroxylase (if available) 4) Pituitary gland: LH, FSH, testosterone, estradiol, prolactin, +/- pituitary gland magnetic resonance imaging

Discussion

Here, we have presented the most comprehensive review of APS-2 following the use of ICPis therapy in patients with advanced solid tumors. ICPis-induced APS-2 cases are rare. Our study demonstrated that, among the patients receiving ICPis therapy, the rates of thyroid and islet disorders were the highest out of all the rates of endocrinopathies; most patients receiving ICPis therapy experienced 2 of the endocrinopathies involved in APS-2: thyroid disease and T1DM. Patients taking PD-1 inhibitors were more likely to experience APS-2 compared to those taking PD-L1 inhibitors or a combination regimen. Furthermore, the most common causative agent of APS-2 was the PD-1 inhibitor pembrolizumab.

Patients with spontaneous APS-2 have at least 2 of the following 3 endocrinopathies: T1DM, autoimmune thyroid disease, and Addison’s disease (9). Women are more likely to have spontaneous APS-2 than men, and the typical age at onset of APS-2 is from adolescence to adulthood. Testing for autoantibodies may be helpful in assessing the disease risk. In contrast, in our review, the median age of ICPis-induced APS-2 was 62 years. Men may be more likely to have ICPis-induced APS-2 than women, as indicated in this study. As in spontaneous APS, organ-specific autoantibodies often accompany the autoimmune manifestations in patients taking ICPis, although their predictive ability and time course is unclear. In genetic studies of spontaneous APS-2, genetic associations mostly involved the major histocompatibility complex (9). Furthermore, the heterozygous HLA-DR4-DQ8/ HLA-DR3-DQ2 genotype increased the risk of Addison’s disease, T1DM, and autoimmune thyroid diseases (35-37). The concurrence of more than one endocrinopathy may result from genetic susceptibility leading to loss of tolerance to multiple tissues. In our review, 71.4% (10/14) of the patients had a high-risk HLA allele (HLA-DR4). This is higher than the rate of HLA-DR4 in patients with spontaneous APS-2 (35.2%) (38), but similar to the rate in patients with PD-1 inhibitor-induced diabetes (76% in a previous study) (39). Based on data from the reviewed literature, we believe that patients with a high-risk HLA allele have an increased risk of ICPis-induced APS-2.

The PD-1/PD-L1 pathway is a key regulator in T-cell activation and peripheral tolerance. PD-1 is generally not expressed on native T cells but rather on chronically activated T cells in peripheral tissues, particularly CD8⁺ T cells. PD-L1 is upregulated by inflammatory cytokines in a wide variety of tissues, including many tumors (40, 41). By binding to its ligands (PD-L1 and PD-L2), PD-1 transmits a negative signal in T cells, thereby inhibiting the immune response (42). When the PD-1 pathway is blocked, not only do T cells targeting cancer survive, but autoreactive T cells also survive, causing autoimmune diseases. Because of the disinhibition of autoreactive T cells, they can survive and destroy the target tissues, such as islets.

The most common treatment regimen in this review was PD-1 inhibitor monotherapy. The differences in the rates of APS-2 patients taking PD-1 inhibitor monotherapy (76.9%), PD-L1 inhibitor monotherapy (7.7%), or a combination of CTLA-4 inhibitors and PD-1 inhibitors (15.4%) suggest that the PD-1 or PD-L1 axes and CTLA-4 have distinct importance in the maintenance of immune tolerance to the thyroid, islets, and adrenal glands. It is still unclear why there were higher rates of APS-2 patients taking PD-1 inhibitor monotherapy compared with those taking PD-L1 inhibitor monotherapy or a combination of CTLA-4 inhibitors and PD-1 inhibitors. However, previous clinical evidence hints at this stronger association between PD-1 inhibitors and APS-2. Previous research found that among patients taking ICPis monotherapy, the incidence of thyroid disease was higher in those treated with anti-PD-1 agents (4) compared with those taking a PD-L1 inhibitor or ipilimumab (a CTLA-4 inhibitor). In addition, most cases of ICPis-induced T1DM have been observed in patients treated with PD-1 inhibitors (4, 43). It is not surprising, therefore, that treatment with PD-1 inhibitors has led to polyendocrine autoimmunity.

The mechanisms behind these observations and their relationships are not yet understood. One study demonstrated that normal thyroid tissue expresses both PD-L1 and PD-L2 (44). Thus, one hypothesis is that PD-L2 blockade is involved in the pathogenesis of thyroid disease (4). Second, a previous study found increased expression of PD-L1 on β-cells from nonobese diabetic (NOD) mice during the progression of autoimmune diabetes (45). PD-L1 is also specifically upregulated on the β-cells of patients with T1DM (46). T-cell activation by PD-1 blockade can cause β-cell destruction, and research has indicated that there is a substantial reduction in the expression of PD-1 on CD4+ T cells in T1DM patients relative to in healthy controls (47).

Notably, most patients who developed ICPis-induced APS-2 (16/21, 76.2%) had a partial or complete antitumor response. The patients taking ICPis therapy mainly had melanoma (14/26, 53.8%). ICPis therapy leads to prolonged progression-free survival and overall survival, with response rates of 45% for melanoma patients taking PD-1 inhibitor monotherapy and 60% for patients taking a combination of a CTLA-4 inhibitor and a PD-1 inhibitor (48). Our results indicate an association between APS-2 and improved clinical outcomes. However, Freeman-Keller et al did not observe a noteworthy difference in survival in melanoma patients with endocrinopathies irAEs treated with nivolumab (a PD-1 inhibitor) (49). As this information is highly clinically relevant, further validation is required and it would be of great interest to study this in larger prospective trials.

Another highlight in our article is that we propose a screening protocol for APS-2. Several consensus-based protocols have previously been suggested, according to which patients should undergo an assessment to identify any autoimmune endocrine diseases both prior to starting ICPis therapy and at follow-up (33, 34). However, no consensus has emphasized the screening and diagnosis of APS-2 (involving multiple autoimmune endocrine diseases). In our opinion, a step-by-step strategy involving a combination of education, monitoring, screening for other ICPis-induced autoimmune diseases after diagnosis of an initial ICPis-induced autoimmune disease, diagnosis and treatment, and follow-up should be used in clinical practice related to APS-2. The risk of appearance of an endocrinopathy is greater at the start of treatment, thus closer monitoring is necessary over the first 6 months, followed by regular monitoring over the next (second) 6 months and less frequent monitoring thereafter. Since the clinical manifestations are often nonspecific, the grades of ICPis-induced hypophysitis or primary adrenal insufficiency are often 3 or 4 at diagnosis, so close monitoring is necessary. Therefore, in addition to monitoring fasting glucose levels (or random plasma glucose levels) and blood electrolytes, we also recommend evaluation of thyrotropin (thyroid-stimulating hormone), thyroxine, adrenocorticotropic hormone, and cortisol levels during each course of treatment during the first 6 months and every 2 courses from months 6 to 12. Also, this advice is based on the French Endocrine Society Guidance on endocrine side effects of immunotherapy (33). ICPis-induced T1DM often presents with DKA. If it is not promptly recognized, this poses a danger to the patient. Most patients with T1DM have mildly to moderately raised HbA1c at diagnosis and to diagnose T1DM early we recommend routine HbA1c measurement, in addition to glucose levels, in patients. Clotman et al also recommend routine measurement of HbA1c and blood glucose levels at each administration of ICPis (43).

The strengths of this study include that it is the most comprehensive review of APS-2 and that we propose, for the first time, an APS-2 screening strategy for patients receiving ICPis therapy. Second, the examined information includes additional potential risk factors potentially associated with the development of endocrine irAEs from the case reports examined, including preexisting endocrine or autoimmune diseases and HLA genotyping. Therefore, we recommend extra vigilance for patients with a history of autoimmune disease and in the screening strategy, we suggest that HLA genotyping be performed to potentially identify susceptible haplotypes.

Still, our study has several limitations that should be recognized. First, because the data was mainly obtained from the individually published cases, the real prevalence of APS-2 during ICPis therapy may have severely underestimated. Thus, the true risk of these events should be ascertained in prospective studies. Second, 4 cases from the FDA and EMA adverse events pharmacovigilance registries and some additional detailed information, which could contribute to a more comprehensive evaluation, were unavailable.

The increasing use of ICPis will likely change the incidence of ICPis-induced APS, and APS-2 remains to be fully elucidated. Physicians (especially endocrinologists and oncologists) should be familiar with APS-2 induced by these new drugs, particularly PD-1 inhibitors. Based on our results, we recommend monitoring for APS-2 in patients taking ICPis (especially PD-1 inhibitors) if they are diagnosed with a single ICPis-induced autoimmune disease, and patients with the high-risk HLA allele require close monitoring. Monitoring can help prevent illness associated with delayed diagnosis of additional autoimmune diseases. However, further prospective studies are needed to explore the risk factors and biomarkers to better identify individuals at risk and manage them appropriately.

Abbreviations

Abbreviations
 
• APS-2

  autoimmune polyendocrine syndrome type 2

 
• CTLA-4

  cytotoxic T-lymphocyte antigen 4

 
• DKA

  diabetic ketoacidosis

 
• EMA

  European Medicines Agency

 
• FDA

  US Food and Drug Administration

 
• HbA1c

  glycated hemoglobin A1c

 
• HLA

  human leukocyte antigen

 
• ICPi

  immune checkpoint inhibitor

 
• irAE

  immune-related adverse event

 
• PD-1

  programmed death-1

 
• PD-L1

  programmed death-ligand 1

 
• T1DM

  type 1 diabetes mellitus

Acknowledgments

Financial Support: This study was supported by grants from the National Natural Science Foundation of China (number 81700689).

Additional Information

Disclosure Summary: The authors have nothing to disclose.

Data Availability

All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.

Author notes

These authors contributed equally to this study.