Ectopic rabphilin-3A expression in lung adenocarcinoma associated with paraneoplastic arginine vasopressin deficiency during anti-PD-L1 therapy: a case report

Abstract

Immune checkpoint inhibitors (ICIs) can cause endocrine immune-related adverse events affecting multiple endocrine organs, including the pituitary, yet arginine vasopressin deficiency (AVP-D) remains rare. Lymphocytic infundibuloneurohypophysitis, the principal cause of AVP-D, is an autoimmune inflammatory disorder of the posterior pituitary and stalk in which anti-RPH3A antibodies serve as a specific marker, and RPH3A as a pathogenic autoantigen, yet its relationship to cancer and ICI therapy remains unclear. We report a 63-year-old man with lung adenocarcinoma treated with carboplatin, paclitaxel, bevacizumab, and atezolizumab who developed AVP-D without obvious anterior pituitary dysfunction. Serum testing demonstrated the presence of anti-RPH3A antibodies. Lung tumor specimens showed ectopic RPH3A expression with prominent CD8⁺ T-cell infiltration by immunohistochemistry and in situ hybridization chain reaction. Given prior evidence that RPH3A is a pathogenic autoantigen in AVP neurons, these findings support a paraneoplastic mechanism in which PD-L1 blockade unleashes RPH3A-reactive T cells primed against ectopic tumor expression, leading to collateral injury of AVP neurons and consequent AVP-D.

Insights

Immune checkpoint blockade with anti-PD‑L1 may unmask paraneoplastic arginine vasopressin deficiency (AVP‑D) by unleashing RPH3A‑reactive T cells primed against ectopic RPH3A expression in lung adenocarcinoma, leading to collateral injury of AVP neurons.

Introduction

Immune checkpoint inhibitors (ICIs) have transformed the treatment of advanced malignancies, but can cause immune-related adverse events (irAEs) affecting multiple endocrine organs, including the pituitary.¹ ICI-induced hypophysitis typically results in secondary adrenal insufficiency and/or central hypothyroidism, whereas ICI-induced arginine vasopressin deficiency (AVP-D; formerly central diabetes insipidus) is rare, with only a few cases reported.² Moreover, the mechanisms underlying these endocrine irAEs remain largely unclear.

Autoimmune-mediated paraneoplastic syndromes are immune-driven disorders in which antitumor immune responses cross-react with normal tissues. In such cases, ectopic expression of neuronal or endocrine antigens by tumor cells can induce autoantibody production and/or cytotoxic T-cell activation, leading to tissue injury.³ In this context, recent mechanistic work has proposed that a subset of PD-1/PD-L1-related hypophysitis represents a paraneoplastic phenomenon: tumors ectopically expressing adrenocorticotropic hormone (ACTH) prime antigen-specific T-cell responses, which are subsequently amplified by checkpoint blockade and redirected toward the pituitary.⁴ These observations provide a conceptual bridge between classical paraneoplastic syndromes and ICI-induced endocrine irAEs.

Lymphocytic infundibuloneurohypophysitis (LINH), the principal cause of AVP-D, is an autoimmune inflammatory disorder that selectively targets the posterior pituitary and pituitary stalk, leading to inflammation and destruction of AVP-producing neurons.⁵ In LINH, serum anti-RPH3A antibodies have been established as a highly specific diagnostic marker.⁶ In addition, experimental mouse models have demonstrated that immunization with RPH3A induces autoimmune neurohypophysitis through RPH3A-specific T-cell responses, supporting RPH3A as a pathogenic autoantigen.⁷

Here, a unique case is described of a patient with lung adenocarcinoma treated with carboplatin, paclitaxel, bevacizumab, and the anti–PD-L1 antibody atezolizumab who developed AVP-D without overt anterior pituitary dysfunction. The case is characterized by positivity for anti-RPH3A antibodies and ectopic RPH3A expression with prominent CD8⁺ T-cell infiltration in lung adenocarcinoma tissue, providing integrated evidence for an RPH3A-directed paraneoplastic mechanism of ICI-unmasked AVP-D.

Methods

The complete experimental methods are described in online supplemental materials.

Case presentation

The patient’s clinical course is summarized in figure 1A. A 63-year-old man was diagnosed with lung adenocarcinoma (figures 1B and 2A). He had a history of type 2 diabetes mellitus but had discontinued medical follow-up and remained untreated for the subsequent 5 years. He was admitted for poor glycemic control, during which an increase in the size of the lung mass was noted (figure 1C). A second biopsy confirmed lung adenocarcinoma. The disease was staged as cT3N0M1a (stage IVA). The tumor was negative for EGFR mutations and ALK rearrangements, with no other actionable driver mutations identified. The PD-L1 tumor proportion score was 30%. Baseline laboratory studies prior to treatment initiation were unremarkable except for elevated HbA1c. Combination therapy with carboplatin, paclitaxel, bevacizumab, and atezolizumab was initiated, and three cycles were completed over 2 months. The best overall response to first-line chemo-immunotherapy was a partial response (PR). After three cycles of first-line chemo-immunotherapy, interstitial pneumonitis developed (figure 1D).

Figure 1. CT and MRI of the patient. (A) Timeline of the clinical course. (B–D) CT images of the lungs obtained at the time of the initial diagnosis of lung cancer (B), 5 years after the initial diagnosis (C), and after three cycles of chemotherapy (D). White arrows indicate the tumor, and red arrows indicate pneumonitis. (E) T1-weighted MRI of the pituitary gland at the time of AVP-D diagnosis. The white arrow indicates the pituitary gland. AVP-D, arginine vasopressin deficiency; ICI, immune checkpoint inhibitor; mo, month; PD, progressive disease; PR, partial response; wk, week; yr, year.

Figure 2. H&E staining, immunohistochemistry (IHC), and in situ hybridization chain reaction (HCR) of lung adenocarcinoma tissue, and detection of anti-RPH3A antibodies by western blotting. (A–D) Lung adenocarcinoma tissue from the initial diagnostic biopsy. H&E staining (A), IHC for RPH3A (B), in situ HCR for RPH3A (C), and IHC for CD8 (D). Scale bar, 50 µm. (E–H) Third lung biopsy specimen obtained at the time of pneumonitis diagnosis. H&E staining (E), IHC for RPH3A (F), in situ HCR for RPH3A (G), and IHC for CD8 (H). Scale bar, 50 µm. (I) Quantitative analysis of in situ HCR signals in the three sequential biopsy specimens. Data are presented as RPH3A transcripts per 1000 pixels². (J) Detection of anti-RPH3A antibodies by western blotting. Recombinant full-length human RPH3A expressed in HEK293FT cells (RPH3A+HEK293 FT, left lane in each case) or negative control (HEK293FT, right lane in each case) was probed with sera from the present patient and from a patient with LINH. Recombinant full-length human RPH3A expressed in HEK293FT cells was also probed with an anti-V5 antibody as a positive control (anti-V5 antibody) in the first lane from the left. Serum from the present patient and a patient with LINH reacted with only recombinant full-length human rabphilin-3A (RPH3A+HEK293 FT lane). LINH, lymphocytic infundibuloneurohypophysitis.

The patient underwent three lung biopsies during the clinical course: the initial diagnostic biopsy (figure 2A–D), a second biopsy approximately 5 years later because of tumor progression (online supplemental figure A), and a third biopsy performed during the evaluation of interstitial pneumonitis (figure 2E–H). Immunohistochemical staining and in situ hybridization chain reaction (HCR) analyses for RPH3A were performed retrospectively using archived biopsy specimens obtained at three different time points (figure 2B,C,F,G, online supplemental figure A). A negative control for in situ HCR targeting RPH3A, performed using lung adenocarcinoma tissue obtained from a different patient, showed almost no punctate signals (online supplemental figure B–E). Compared with the initial diagnostic biopsy specimen, quantitative image analysis of in situ HCR signals demonstrated increased RPH3A expression in the second biopsy specimen obtained approximately 5 years later, with the mean in situ HCR puncta density per nuclear ROI increasing from 3.38 puncta per 1000 pixels² in the first biopsy to 6.23 in the second biopsy and remaining at a similar level in the third biopsy (6.66 puncta per 1000 pixels²) (figure 2I).

Seven months after the initiation of chemotherapy, he sustained an avulsion fracture of the left lesser trochanter secondary to bone metastasis and underwent open reduction and internal fixation. Postoperatively, polyuria became evident, with a urine output of 2600 mL over 12 hours. A detailed medical interview revealed that he had been consuming approximately 5 L of water per day prior to admission. Laboratory evaluation was consistent with AVP-D, showing elevated plasma osmolality (301 mOsm/kg), and inappropriately low urine specific gravity (1.003) and plasma AVP concentration (0.6 pg/mL) (online supplemental table 1). Findings did not suggest thyroid or adrenal insufficiency (online supplemental table 1). MRI demonstrated a loss of the normal T1-weighted high-intensity signal in the posterior pituitary, without pituitary enlargement or stalk thickening (figure 1E). Serum anti-RPH3A antibodies were detected by western blotting (figure 2J).

Intranasal desmopressin was initiated and titrated up to 20 µg/day, resulting in complete resolution of symptoms. The patient did not receive any second-line systemic anticancer therapy during the subsequent clinical course. Best supportive care was subsequently continued, but the patient died 1 month after the open reduction and internal fixation surgery.

Discussion

We previously identified anti-RPH3A antibody as a useful diagnostic marker for LINH, detectable in 76% (22/29) of patients—including all biopsy-proven cases—while absent in 34 non-LINH sellar/suprasellar lesions.⁶ Furthermore, a preclinical mouse model demonstrated that immunization with RPH3A induces autoimmune LINH characterized by CD3⁺/CD8⁺ T-cell infiltration, polyuria, and RPH3A-specific T-cell responses, supporting RPH3A as a pathogenic autoantigen.⁷

RPH3A, identified as an effector of the small GTPase RAB3A, is a key regulator of secretory vesicle trafficking and is predominantly expressed in the brain.⁸ In lung adenocarcinoma, the Human Protein Atlas shows no detectable RPH3A protein expression, and analyses of TCGA and GTEx datasets reveal comparable RPH3A mRNA levels between tumor and normal lung tissues, suggesting that RPH3A is not generally dysregulated at either the transcriptional or protein level in this context.⁹ Immunohistochemical staining and in situ HCR analyses for RPH3A were performed retrospectively using archived biopsy specimens obtained during the patient’s clinical course at three different time points. In the present case, RPH3A expression was clearly demonstrated in lung adenocarcinoma tissue by immunohistochemistry and in situ HCR, indicating ectopic expression. Furthermore, RPH3A expression appeared to be higher than that observed 5 years earlier, accompanied by prominent infiltration of CD8⁺ T cells within the tumor. These findings suggest that ectopic RPH3A expression may have elicited an RPH3A-directed cytotoxic T-cell response. Our proposed mechanism is further supported by our previous preclinical mouse model, in which immunization with RPH3A induced autoimmune LINH characterized by CD3⁺ and CD8⁺ T-cell infiltration into the neurohypophysis and supraoptic nucleus, polyuria with hypotonic urine, and RPH3A-specific T-cell responses.⁷ The increased RPH3A expression together with marked CD8⁺ T-cell infiltration observed in the present case is consistent with these experimental findings and further supports the biological plausibility of an immune-mediated cross-reactive mechanism. Given that RPH3A is physiologically expressed in AVP neurons,⁶ direct immune activation by checkpoint blockade alone would be expected to produce AVP deficiency more frequently. The rarity of this complication raises the possibility that additional antigenic priming, such as ectopic tumor expression of RPH3A, may be required.

Although transbronchial lung biopsy did not provide histopathological confirmation of pneumonitis, the diagnosis was made based on the overall clinical course and serial radiological findings after exclusion of alternative causes, including carcinomatous lymphangitis and opportunistic infections. Immune-related pneumonitis was considered the most likely etiology based on the temporal clinical course and radiological findings, whereas taxane-induced pneumonitis was considered less likely. Notably, although the biopsy was performed for the evaluation of pneumonitis, it unexpectedly contained a microscopic metastatic adenocarcinoma lesion, providing a unique opportunity to retrospectively investigate longitudinal changes in RPH3A expression across sequential biopsy specimens.

Although uncommon, several reports have suggested that anti-CTLA-4 antibody-induced AVP-D and anti-PD-1/PD-L1 antibody-induced AVP-D exhibit distinct clinical features.² Anti-CTLA-4 antibodies appear to cause AVP-D only in the setting of panhypophysitis, whereas anti-PD-1/PD-L1 antibodies induce AVP-D as an isolated endocrine irAE mirroring the clinical pattern observed in the present patient. Consistent with this, the mechanism of anti-CTLA-4-related pituitary irAE has been attributed to CTLA-4 expression on anterior pituitary cells, which permits direct binding of the therapeutic antibody and subsequent complement-dependent cytotoxicity.¹⁰ In contrast, the pathogenesis of anti-PD-1/PD-L1 antibody-induced pituitary irAEs remains largely unclear.

In our case, anti-PD-L1 therapy was followed by anti-RPH3A antibody–associated AVP-D. Anterior pituitary function was largely preserved, with normal pituitary-adrenal and pituitary-thyroid function. Only a mildly decreased serum IGF-1 level and a mildly elevated prolactin level were observed. Therefore, given the ectopic expression of RPH3A in the lung adenocarcinoma, we speculate that RPH3A-reactive T cells—normally suppressed by peripheral tolerance mechanisms—were unleashed by PD-L1 blockade, leading to recognition of RPH3A on tumor cells and collateral damage of AVP-producing neurons. The anti-RPH3A antibody in this context likely represents a secondary marker of neuronal destruction rather than a pathogenic mediator.⁷ A similar paraneoplastic autoimmune hypophysitis mechanism has been suggested for a subset of PD-1/PD-L1-related anterior pituitary hypophysitis leading to adrenal insufficiency with ectopic ACTH expression and corticotroph-directed autoimmunity.⁴ In contrast to these corticotroph-predominant anterior pituitary cases, our report highlights a posterior pituitary–dominant phenotype. An important limitation of this case is the lack of serum samples obtained before the initiation of immunotherapy. Therefore, we could not determine whether anti-RPH3A antibodies were already present before PD-L1 blockade or emerged during the subsequent clinical course. Nevertheless, our case uniquely combines serological evidence of anti-RPH3A autoimmunity with direct demonstration of ectopic RPH3A expression and dense CD8⁺ T-cell infiltration within the lung adenocarcinoma, thus providing integrated clinical, serological, and histopathological support for a paraneoplastic mechanism of ICI-unmasked AVP-D.

Conclusion

We report a case of AVP-D associated with lung adenocarcinoma expressing ectopic RPH3A, likely triggered by anti-PD-L1-enhanced autoimmune cross-reactivity. This case highlights a novel paraneoplastic mechanism linking ectopic antigen expression, immune checkpoint blockade, and AVP-D.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.