Immune checkpoint inhibitor-associated severe hyponatremia: evidence mapping of SIADH-like phenotypes and endocrine immune-related adverse events

Abstract

Objective

Severe or symptomatic hyponatremia during immune checkpoint inhibitor (ICI) therapy may reflect syndrome of inappropriate antidiuresis (SIADH), endocrine immune-related adverse events (irAEs), cancer-related factors, or mixed mechanisms. We mapped the evidence and reclassified diagnostically extractable cases to distinguish SIADH-like phenotypes from endocrine irAEs.

Methods

PubMed, Embase, Web of Science Core Collection, Scopus, and the Cochrane Library were searched through May 24, 2026. Reports were assigned to A1 case-level, A2 population-level, or A3 diagnostic-framework layers. A1 summaries used patient/case denominators after a report-versus-patient audit. Diagnostic categories were reviewer-derived. Reporting completeness was described using six prespecified domains. We performed source-label and full-length-only sensitivity analyses.

Results

All 194 reports sought for retrieval were assessed, and 146 were included: 127 A1 reports describing 144 patients/cases, 16 A2 reports, and 3 A3 reports. Strict classification identified 127/144 (88.2%) confirmed/probable adrenal-axis irAEs, 1/144 (0.7%) confirmed/probable SIADH, 3/144 (2.1%) SIADH-like phenotypes with incomplete endocrine exclusion, 2/144 (1.4%) thyroid-related cases, and 11/144 (7.6%) mixed, confounded, or non-endocrine mechanisms. Ten patients/cases carried an explicit source-level SIADH label; strict review classified six as adrenal-axis irAEs, one as SIADH, and three as SIADH-like with incomplete endocrine exclusion. After exclusion of 81 abstract/database-only A1 reports, 46 full-length reports described 59 patients: 52 adrenal-axis irAEs, 1 SIADH, and 6 mixed/confounded or non-endocrine cases.

Conclusion

Published classifiable A1 cases were dominated by adrenal-axis irAEs. Strictly supported SIADH occurred but was rare in this selected evidence base, and the direction of findings persisted in the full-length-only analysis. These proportions are not incidence estimates. The proposed algorithm supports diagnostic sequencing and transparent reporting; it is not a stand-alone practice guideline.

Introduction

Immune checkpoint inhibitors are standard therapy for many cancers, but endocrine irAEs may present abruptly and overlap with cancer progression, infection, treatment toxicity, or paraneoplastic syndromes. These toxicities can require urgent recognition, hormone replacement, treatment modification, and coordination among oncology, endocrinology, and acute-care teams (1, 2, 3, 4, 5, 6, 7, 8, 9, 10).

Although this review focuses on endocrine phenotypes, checkpoint-mediated immune regulation is not confined to endocrine tissues. Ocular checkpoint biology provides a tissue-specific example in an immune-privileged organ. Separately, evidence for sustained-release intravitreal dexamethasone implants in the broader, non-ICI-specific uveitis literature illustrates locally directed anti-inflammatory management (11, 12). These contextual examples do not inform attribution of hyponatremia and were not included in the A1, A2, or A3 evidence layers.

Hyponatremia is common in cancer care, whereas severe or symptomatic hyponatremia has immediate clinical importance (13, 14, 15, 16, 17, 18, 19). Competing causes include low solute intake, gastrointestinal loss, diuretics, infection, pulmonary or central nervous system disease, tumor progression, and paraneoplastic SIADH. ICI exposure adds adrenal insufficiency, isolated adrenocorticotropic hormone (ACTH) deficiency, hypophysitis, and thyroid dysfunction to this differential.

Adrenal-axis failure can increase vasopressin-mediated water retention and produce a SIADH-like biochemical phenotype (13, 14, 20). A patient with fatigue, nausea, anorexia, dizziness, hypotension, or confusion may therefore receive an early SIADH label before cortisol, ACTH, thyroid-stimulating hormone (TSH), and free thyroxine (FT4) have been adequately assessed. Thyroid dysfunction and mixed etiologies further complicate attribution.

The evidence is fragmented across case reports, conference abstracts, cohorts, pharmacovigilance studies, systematic reviews, and diagnostic frameworks. These sources answer different questions and should not be pooled as if they were interchangeable. We therefore conducted a systematic scoping review with evidence mapping and diagnostic reclassification. We aimed to map case-level, population-level, and framework evidence; audit report and patient denominators; quantify reporting completeness; compare source SIADH labels with strict review categories; and develop cautious clinical decision support. We did not aim to estimate incidence, perform meta-analysis, or issue a clinical-practice guideline.

Methods

Review design and reporting framework

We performed a systematic scoping review with evidence mapping and diagnostic reclassification, informed by PRISMA-ScR and established scoping review methodology (21, 22, 23). The phenotype-driven question was how clinically meaningful hyponatremia after ICI exposure had been reported and whether extractable information supported SIADH, endocrine irAEs, or other/mixed mechanisms. Heterogeneous evidence types were mapped rather than statistically pooled.

Three evidence layers were used (Table 1). A1 comprised diagnostically extractable case-level evidence. A2 comprised population-level evidence in which hyponatremia or an electrolyte abnormality was an analyzable outcome. A3 comprised evidence directly supporting diagnostic classification. PRISMA and layer counts used reports; A1 clinical summaries used individual patients/cases.

Table 1.

Evidence-layer definitions and corrected denominators.

Layer	Inclusion criteria	Key exclusions	Denominator	Analytical purpose	Contribution to conclusions
A1	ICI exposure; clinically meaningful hyponatremia; extractable patient-level diagnostic or treatment information	Mild/asymptomatic-only sodium findings; no patient-level diagnostic/treatment details; no ICI exposure	127 reports/144 patients	Diagnostic attribution and reporting completeness	Primary strict distribution and patient-level sensitivity analyses
A2	Analyzable aggregate hyponatremia or electrolyte outcome	Routine safety-table mention; no analyzable electrolyte outcome	16 reports	Frequency, severity, prognosis, and signal context	Context only; no A1 patient reclassification or denominator contribution
A3	Direct support for SIADH/endocrine diagnostic reasoning	Background content without direct diagnostic-framework relevance	3 reports	Support classification concepts and terminology	Informs rules; contributes no patient events

Report counts are mutually exclusive. PRISMA uses reports; A1 diagnostic summaries use audited patients/cases.

Search strategy and information sources

PubMed, Embase, Web of Science Core Collection, Scopus, and the Cochrane Library were searched for English-language records from 2011 through May 24, 2026. Database platforms, complete search strings, dates, limits, and export counts are reproduced in Supplementary Table S1 (see section on Supplementary materials given at the end of the article). Database exports were merged and deduplicated using PMID, DOI, title similarity, year, and first-author information. The search identified 7,877 records; 3,609 duplicates were removed, leaving 4,268 records for title/abstract screening.

Conference abstracts and database records were eligible only when a complete published abstract or sufficiently detailed retrievable record had stable bibliographic metadata and independently met the criteria for an evidence layer. These sources were explicitly labeled and were not treated as equivalent to full-length articles. No information beyond the retrievable source was inferred or imputed. A sensitivity analysis excluded all abstract/database-only A1 reports.

Eligibility, retrieval, and eligibility-stage disposition

A1 required cancer-patient exposure to an ICI, explicit hyponatremia or a serum sodium value, clinically meaningful severity, and extractable patient-level diagnostic or treatment information. Clinical significance included sodium ≤125 mmol/L, grade 3/4 or otherwise severe hyponatremia, symptoms, emergency/hospital/intensive-care treatment, adrenal crisis, or active sodium- or cause-directed treatment. A2 required an analyzable population-level hyponatremia/electrolyte outcome; routine safety-table mentions were excluded. A3 required direct relevance to SIADH or endocrine diagnostic reasoning and contributed no patient events.

All 194 reports sought for retrieval were directly retrievable or otherwise verifiable and underwent eligibility assessment. Two reports initially recorded as not retrieved were located during the final revision-stage reconciliation: one met A1 criteria and one met A2 criteria. A third report previously excluded for non-retrieval was re-verified from its official publisher extract and met A1 criteria. Forty-eight reports were excluded: no extractable patient-level diagnostic/treatment information (n = 18), duplicate/companion reports (n = 12), source-specific ineligibility (n = 5), no ICI exposure (n = 6), routine safety-table reporting (n = 4), source mismatch (n = 1), wrong study design (n = 1), or no explicit patient-level hyponatremia/low-sodium outcome (n = 1). All eligibility-stage dispositions are listed in Supplementary Table S5.

Revision-stage source audit and denominator control

During revision, all 194 sought reports were audited against the prespecified layer definitions, retrievable sources, identifiers, and report-versus-patient structure. This audit corrected under-inclusion, retrieval status, source mismatches, companion duplicates, category assignments, and patient-level extraction inconsistencies in the initial extraction. It did not broaden the clinical significance threshold or add post-search records. A decision log and source-level reference audit were retained. The corrected synthesis comprised 146 included reports: 127 A1, 16 A2, and 3 A3.

A1 report and patient denominators were separated. Multi-patient reports were split only when the number of qualifying patients was confirmable; report-level details were not assigned to individual patients without support. Missing variables remained not reported. Supplementary Table S6 documents report-to-patient mapping.

Data extraction and reporting completeness

For each A1 patient/case, we extracted ICI exposure, cancer type, sodium nadir, symptoms and care setting, serum osmolality, urine osmolality, urine sodium, adrenal- and thyroid-axis information, source diagnosis, treatment, competing factors, and outcome when available. Six reporting domains were counted: sodium nadir, serum osmolality, urine osmolality, urine sodium, thyroid-axis information, and adrenal-axis information. These domains pragmatically represented severity, confirmation of hypotonicity, renal water/sodium handling, and exclusion of the two principal endocrine mimics. Equal weighting reflected a simple availability count and did not imply equal clinical importance. The low (0–2), moderate (3–4), and high (5–6) strata were descriptive groupings chosen to make the distribution interpretable. The index was not validated, was not a risk-of-bias instrument, and was not used to exclude reports or weight diagnostic categories.

Diagnostic reclassification and sensitivity analyses

Final categories were confirmed/probable adrenal-axis irAEs, confirmed/probable SIADH, SIADH-like phenotype with incomplete endocrine exclusion, thyroid-related hyponatremia, and mixed/confounded or non-endocrine mechanisms. Confirmed/probable SIADH required a compatible hypotonic phenotype, supportive urine studies where available, and adequate exclusion of adrenal- and thyroid-axis causes. A source-author SIADH label alone was insufficient. Response to fluid restriction or tolvaptan was recorded as treatment information rather than diagnostic proof. Mixed mechanisms were retained when supported.

Because reclassification was reviewer-derived, source-author diagnostic labels were preserved separately. We compared all explicit source-level SIADH labels with strict categories. We also repeated the A1 diagnostic distribution after excluding all conference-abstract and database-record sources, retaining full-length reports only. No inferential comparison or meta-analysis was performed.

Reliability and methodological limitations

One revision-stage classifier applied rules fixed before the full re-audit and checked each decision against the source-level extraction and identifier record. A second independent classifier was not used; disagreements and inter-rater agreement were therefore not applicable or quantifiable, and the rules were not externally piloted. Automated denominator assertions were used as a separate arithmetic control. Ambiguous records were retained in conservative categories, and source-specific uncertainty is visible in Supplementary Table S2. This limitation is considered when interpreting all reclassified proportions.

Results

Study selection and evidence layers

After the removal of 3,609 duplicates, 4,268 records were screened. All 194 reports sought were retrieved or otherwise directly verified and assessed. Forty-eight were excluded, and 146 reports entered the evidence map: 127 A1 reports, 16 A2 reports, and 3 A3 reports (Fig. 1; Table 1). The A1 reports described 144 individual patients/cases. Eighty-one A1 reports (85 patients) were conference abstracts or database records; 46 full-length reports described 59 patients. Included A3 reports are listed in references (24, 25, 26). Included A2 reports are listed in references (27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42). Included A1 reports are listed in references (43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169).

Figure 1.

PRISMA-ScR flow diagram. The revised source audit identified 7,877 records, removed 3,609 duplicates, screened 4,268 records, sought and assessed all 194 reports, excluded 48 reports, and included 146 reports: 127 A1, 16 A2, and 3 A3. No report remained not retrieved. Abstract/database-only evidence is explicitly labeled and is not treated as full-length evidence.

A1 diagnostic distribution

Strict review classification identified confirmed/probable adrenal-axis irAEs in 127/144 patients/cases (88.2%), confirmed/probable SIADH in 1/144 (0.7%), SIADH-like phenotypes with incomplete endocrine exclusion in 3/144 (2.1%), thyroid-related hyponatremia in 2/144 (1.4%), and mixed/confounded or non-endocrine mechanisms in 11/144 (7.6%) (Fig. 2; Table 2). These values describe the selected published A1 evidence and are not incidence estimates.

Figure 2.

Evidence map and strict diagnostic distribution. A1 included 127 reports describing 144 patients/cases; A2 included 16 reports and A3 included 3 reports. The A1 categories were adrenal-axis irAEs (127/144), confirmed/probable SIADH (1/144), SIADH-like incomplete endocrine exclusion (3/144), thyroid-related hyponatremia (2/144), and mixed/confounded or non-endocrine mechanisms (11/144). The figure also displays six reporting domains, the prespecified completeness strata, and the source-author-label sensitivity analysis.

Table 2.

A1 strict diagnostic distribution and sensitivity analyses.

Diagnostic category	All A1 patients/cases, n (%)	Full-length-only, n	Source-labeled SIADH, n
Confirmed/probable adrenal-axis irAEs	127/144 (88.2)	52/59	6/10
Confirmed/probable SIADH	1/144 (0.7)	1/59	1/10
SIADH-like phenotype with incomplete endocrine exclusion	3/144 (2.1)	0/59	3/10
Thyroid-related hyponatremia	2/144 (1.4)	0/59	0/10
Mixed/confounded or non-endocrine mechanisms	11/144 (7.6)	6/59	0/10

The full-length-only analysis excluded 81 abstract/database-only A1 reports (85 patients). Source-labeled SIADH is shown separately from review classification. Proportions are descriptive and are not incidence estimates.

Ten patients/cases had an explicit source-level SIADH label. Under the strict rule, six were reclassified as adrenal-axis irAEs because adrenal-axis abnormalities were present, one retained confirmed/probable SIADH after hypotonicity, urine findings, endocrine exclusion, and fluid-restriction response were verified, and three remained SIADH-like with incomplete endocrine exclusion. This comparison demonstrates why source terminology and review classification were kept separate.

Full-length-only sensitivity analysis

Excluding all 81 abstract/database-only A1 reports left 46 full-length reports describing 59 patients. The strict distribution remained directionally similar: 52/59 confirmed/probable adrenal-axis irAEs, 1/59 confirmed/probable SIADH, and 6/59 mixed/confounded or non-endocrine mechanisms. No full-length-only case fell into the thyroid-related or incomplete-endocrine-exclusion category. Thus, abstract-level evidence materially affected the denominator but did not create the adrenal predominance or the single confirmed SIADH finding.

Diagnostic reporting completeness

Sodium nadir was available in 117/144 patients/cases, serum osmolality in 21/144, urine osmolality in 23/144, urine sodium in 23/144, thyroid-axis information in 90/144, and adrenal-axis information in 102/144. Exploratory completeness was low in 70/144, moderate in 54/144, and high in 20/144 (Table 3). The frequent absence of osmolality and urine measurements restricts retrospective confirmation of SIADH, while incomplete endocrine testing restricts confident exclusion of irAEs.

Table 3.

A1 diagnostic reporting completeness.

Domain or score stratum	Reported, n/n	Percentage
Sodium nadir	117/144	81.3%
Serum osmolality	21/144	14.6%
Urine osmolality	23/144	16.0%
Urine sodium	23/144	16.0%
Thyroid-axis information	90/144	62.5%
Adrenal-axis information	102/144	70.8%
Low completeness (0–2 domains)	70/144	48.6%
Moderate completeness (3–4 domains)	54/144	37.5%
High completeness (5–6 domains)	20/144	13.9%

The six-domain score is an exploratory reporting index, not a validated risk-of-bias instrument. Percentages may not total 100 because of rounding.

A2 population-level and A3 framework evidence

The 16 A2 reports comprised cohorts, pharmacovigilance analyses, and evidence syntheses with analyzable electrolyte outcomes. They provided context on frequency, severity, prognosis, reporting patterns, and agent-level signals, but generally lacked the patient-level hormone and urine data required for reclassification. They were therefore not pooled with A1 and did not contribute to the 144-patient diagnostic denominator. The three A3 reports directly informed diagnostic reasoning but contributed no patient events. The analytical roles and findings are detailed in Supplementary Tables S3 and S4.

Clinical decision support

Figure 3 translates the evidence map and established hyponatremia/endocrine principles into a diagnostic sequence. In symptomatic or severe hyponatremia, emergency treatment should not await complete testing. When feasible, without delaying glucocorticoids, serum cortisol and ACTH should be sampled before treatment. Serum and urine osmolality, urine sodium, thyroid tests, volume assessment, medication review, infection evaluation, and cancer context should be integrated. SIADH should be retained only after endocrine causes and competing explanations have been adequately assessed. Mixed mechanisms remain permissible. The figure is decision support and an evidence-reporting aid, not a graded treatment recommendation.

Figure 3.

Clinical decision support for symptomatic or severe hyponatremia during ICI therapy. Emergency treatment should not be delayed for complete testing. Cortisol and ACTH should be sampled before glucocorticoids when feasible without delaying treatment. Serum and urine studies, thyroid testing, competing causes, and mixed mechanisms should be considered. The diagram supports diagnostic sequencing and reporting; it is not a stand-alone treatment guideline.

Discussion

This re-audited evidence map shows that adrenal-axis irAEs dominate published, diagnostically extractable cases of clinically meaningful ICI-associated hyponatremia. Unlike the initial extraction, the corrected dataset contains one case meeting the strict SIADH definition. The defensible conclusion is therefore not that SIADH is absent, but that rigorously supported SIADH is rare within this selected literature and is substantially outnumbered by adrenal-axis disease.

The source-label analysis clarifies a recurrent clinical problem. Six of ten source-labeled SIADH cases had adrenal-axis abnormalities and were reassigned to the adrenal category. Conversely, a single case retained SIADH because hypotonicity, urine findings, endocrine exclusion, and response to fluid restriction formed a coherent diagnostic chain. The distinction is clinically important because glucocorticoid deficiency can mimic SIADH physiology, while fluid restriction alone may be unsafe if adrenal crisis is missed.

The full-length-only analysis strengthens the direction of inference while also exposing the evidence base’s limitations. Abstract/database-only material accounted for 81 of 127 A1 reports. Removing it reduced the A1 population from 144 to 59 patients but retained adrenal predominance and the confirmed SIADH case. This sensitivity analysis does not remove publication bias or heterogeneous case selection; it shows only that the main pattern was not generated solely by abstract-level sources.

Reporting completeness remained poor. Serum and urine osmolality and urine sodium were reported in only a minority of cases. Endocrine testing was more common but still incomplete. The six-domain score was intentionally descriptive; it has not been externally validated, does not measure all diagnostic quality domains, and should not be interpreted as a risk-of-bias scale. Better reports should provide pre-treatment cortisol and ACTH when feasible, TSH and FT4, serum and urine osmolality, urine sodium, volume assessment, concomitant medications, competing causes, timing, and response to cause-specific treatment.

Population-level evidence addresses a different question. Cohorts and pharmacovigilance analyses indicate that hyponatremia and electrolyte disorders occur during ICI therapy, but aggregate data usually cannot determine mechanism. Combining such reports with patient-level cases would create false denominator precision. The layered structure therefore separates signal detection and prognosis from diagnostic attribution.

Figure 3 combines review findings with established clinical principles. The evidence supports early endocrine exclusion and explicit retention of mixed etiologies; it does not directly compare diagnostic algorithms or treatments. Emergency management and hormone-sampling statements are consequently framed as established-principle clinical decision support rather than review-derived graded recommendations.

Limitations

This review has important limitations. Case reports and abstracts are highly selected and subject to publication and language bias. The A1 proportions cannot estimate incidence or comparative risk. Eighty-one A1 reports were abstract/database-only. Diagnostic information was frequently missing, and the classification rules were reviewer-derived. The revision-stage audit corrected source, retrieval, and denominator errors but was not a formal duplicate independent reassessment; inter-rater reliability could not be calculated. Some source records had incomplete bibliographic details. Heterogeneity precluded meta-analysis, and the full-length sensitivity analysis addresses source format rather than all forms of bias. The clinical algorithm was not prospectively validated.

Conclusion

Across 127 A1 reports describing 144 patients/cases, 127 were classified as confirmed/probable adrenal-axis irAEs and one as confirmed/probable SIADH. After the exclusion of abstract/database-only reports, 52/59 full-length cases remained adrenal-axis irAEs and 1/59 remained SIADH. Thus, SIADH can occur during ICI therapy but was rare in the published classifiable evidence, whereas adrenal-axis disease was dominant. These selected-case proportions are not incidence estimates. A safe diagnostic sequence prioritizes emergency care, timely adrenal-axis assessment, thyroid testing, serum and urine studies, and explicit consideration of mixed causes.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Funding

This work was supported by the Intramural Research Fund of Gansu Provincial Hospital under the Science and Technology Innovation Platform Program.

Author contribution statement

SH and YG conceived the study. SH and FG designed the methodology. FG, ZX, and WC performed literature search and screening. FG, ZX, and WC extracted data. SH and YG performed data verification and quality control. SH and FG wrote the original draft. All authors reviewed and edited the manuscript. YG and SH supervised the study. All authors approved the final version of the manuscript.

Ethical approval

Ethical approval was not required because this study was a systematic scoping review of previously published literature. It involved no recruitment of human participants, no intervention, and no collection of new identifiable patient data.

Data availability

All data extracted and analyzed in this review are provided in the supplementary tables. Additional review materials may be made available from the corresponding author upon reasonable request.

AI-assisted language tools

During manuscript preparation and revision, the authors used AI-assisted tools for English-language editing, structural organization, reference-format checking, document production, and deterministic preparation of non-photographic figures. No generative image model was used. AI tools did not generate original study data or independently determine study eligibility or patient diagnoses. All outputs were verified, revised, and approved by the authors, who take full responsibility for the final content.