Case 1

A 37-year-old nulligravid woman presented with a watery vaginal discharge. Cervical biopsy confirmed GAS. A pelvic examination suggested vaginal and parametrial involvement. Magnetic resonance imaging (MRI) revealed an ill-defined 3-cm cervical lesion with anterior vaginal wall thickening, while computed tomography (CT) showed no distant metastasis or significant lymphadenopathy. Serum tumor markers, including carcinoembryonic antigen (CEA), carbohydrate antigen 125 (CA125), and carbohydrate antigen 19-9 (CA19-9), were within normal ranges. The patient was diagnosed with International Federation of Gynecology and Obstetrics (FIGO) 2018 clinical stage IIIA disease. The patient was premenopausal, and surgery was performed in the proliferative phase. The patient underwent abdominal radical hysterectomy with bilateral salpingo-oophorectomy and total vaginectomy. Intraoperative peritoneal cytology was negative.

A macroscopic examination of the hysterectomy specimen showed anterior vaginal wall thickening without a discrete cervical mass (Fig. 1A). Histologically, gastric-type cervical adenocarcinoma diffusely involved the entire cervix and extended to the lower uterine corpus, vaginal wall, and parametrium, as illustrated by serial section mapping (Fig. 1B–C). The histological tumor size was 90 mm, with full-thickness cervical stromal invasion (20/20 mm), lymphovascular space invasion, and pelvic lymph node metastasis. Hematoxylin and eosin (HE) staining demonstrated infiltrative irregular glands lined by mucinous columnar cells with a pale-to-clear cytoplasm and variable nuclear atypia (Fig. 1D). Alcian blue/Periodic acid–Schiff (AB/PAS) staining showed predominantly PAS-positive (Alcian blue–negative) intracellular mucin (Fig. 1E). Tumor glands were positive for gastric markers including HIK1083 and claudin 18 (CLDN18), whereas p16 was negative (Fig. 1F). The final pathological diagnosis was HPV-independent adenocarcinoma, GAS, pT3aN1M0 (UICC 2021), pathological stage IIIC1 (FIGO 2018). The anterior vaginal wall margin was positive. Two months later, a refractory urethral fistula developed, and recurrence was suspected (posterior bladder wall thickening with bilateral inguinal lymphadenopathy); therefore, the patient underwent total cystourethrectomy with ureterocutaneostomy, partial omentectomy, and bilateral inguinal lymphadenectomy. Pathology confirmed GAS in the omentum only. The patient subsequently received whole-pelvic irradiation and remained free of recurrence during the follow-up. Two years after the initial surgery, the patient was transferred for the management of refractory intestinal obstruction.

Fig. 1The alternative text for this image may have been generated using AI.

Gross and histopathological features of gastric-type cervical adenocarcinoma (GAS) in Case 1. (A, B) Gross findings of the radical hysterectomy specimen opened by an anterior longitudinal incision through the cervix and vagina. (A) The endocervical and endometrial surfaces appear smooth without a distinct mass. (B) The area outlined by the yellow dashed line indicates the extent of histologically confirmed tumor involvement: GAS diffusely infiltrates the entire cervix and extends to the lower uterine corpus, vaginal wall, and parametrium. (C) Twelve serial transverse sections of the resected uterus; the yellow dashed line denotes the distribution of the microscopic tumor. (D–F) Representative microscopic findings of GAS in the cervix. (D) Hematoxylin and eosin (HE) staining (200×) shows the infiltrative proliferation of irregular glands lined by mucinous columnar cells with an abundant pale-to-clear cytoplasm and variable nuclear atypia, consistent with GAS. (E) Alcian blue/Periodic acid–Schiff (AB/PAS) staining (200×) shows mainly PAS-positive (neutral) intracellular mucin with weak Alcian blue staining, highlighted as magenta/red. (F) Immunohistochemistry for p16 (200×) was negative in tumor glands. (G–I) The endometrium infiltrated by GAS glands. (G) HE staining (40×) shows GAS glands infiltrating among normal endometrial glands and stroma; the black boxed area is shown at a higher magnification in (H) and (I). (H) HE staining (100×) and (I) claudin 18 (CLDN18) immunohistochemistry (100×) of serial sections show CLDN18-positive GAS glands (black arrows) interspersed between CLDN18-negative normal endometrial glands (red arrows) without any stromal reaction

The morphology of the background endometrium was consistent with the proliferative phase on HE staining (Fig. 1G–H), indicating that “symbiotic” glands were intermingled predominantly within the functional layer. Microscopically, CLDN18-positive GAS glands infiltrated from the myometrium into the endometrium and intermingled with CLDN18-negative normal endometrial glands without disrupting the endometrial structure or any stromal reaction (Fig. 1I). Figure 2A–D shows representative CD8 immunohistochemistry for the four regions. We quantified the density of CD8-positive TILs at 100× magnification (cells/mm2) in 15 hotspot fields per region (n = 15). The endometrial “symbiotic” invasion area showed a lower CD8-positive TILs density (median 151 [interquartile range(IQR) 136.0–162.5]) than the normal endometrium (median 243 [190.0–283.0]), the tumor center of the primary cervical lesion (median 457 [393.0–553.0]), or the invasive margin of the primary cervical lesion (median 274 [232.0–325.0]) (P < 0.001 for all pairwise comparisons after the Bonferroni correction; Fig. 2E). Programmed cell death protein 1 (PD-1) immunostaining revealed PD-1–positive immune cells, and programmed death-ligand 1 (PD-L1) expression assessed by the combined positive score (CPS) was 0 in the “symbiotic” area and normal endometrium, > 10 in the tumor center, and 4 at the invasive margin (Fig. 3A–H).

Fig. 2The alternative text for this image may have been generated using AI.

Comparison of CD8-positive tumor-infiltrating lymphocytes (TILs) across four regions in Case 1. (A–D) Representative CD8 immunohistochemistry (100×) illustrating the distribution of CD8-positive TILs in (A) the normal endometrium (density, 177 cells/mm2), (B) the “symbiotic” area (135 cells/mm2), (C) the tumor center (533 cells/mm2), and (D) the invasive margin (312 cells/mm2). (E) Box-and-whisker plots comparing the density of CD8-positive TILs quantified in 15 hotspot fields per region. Boxes indicate the interquartile range, with the median shown as a horizontal line; whiskers extend to the most extreme values within 1.5× the interquartile range, and dots represent outliers beyond this range. Pairwise comparisons were performed using the Mann–Whitney U test, with the Bonferroni correction for multiple comparisons (Bonferroni-adjusted significance threshold, P < 0.0083 [0.05/6]). The “symbiotic” area showed a significantly lower density of CD8-positive TILs than the other regions (all P < 0.001)

Fig. 3The alternative text for this image may have been generated using AI.

Immunohistochemical staining of programmed cell death protein 1 (PD-1) and programmed death-ligand 1 (PD-L1) across different regions in Case 1. All panels are shown at 200×. (A, B) The normal endometrium stained for (A) PD-1 and (B) PD-L1; combined positive score (CPS) = 0. (C, D) The “symbiotic” area stained for (C) PD-1 and (D) PD-L1; CPS = 0. (E, F) The tumor center stained for (E) PD-1 and (F) PD-L1; CPS > 10. (G, H) The invasive margin stained for (G) PD-1 and (H) PD-L1; CPS = 4

Case 2

A 37-year-old nulligravid woman with no history of sexual intercourse presented with abnormal genital bleeding. Cervical biopsy confirmed adenocarcinoma with a negative high-risk HPV test. A pelvic examination suggested right parametrial involvement. CA19-9 was markedly elevated (2651.8 U/mL), whereas CEA and CA125 were not. MRI revealed an ill-defined 4-cm cervical tumor, and CT showed no distant metastasis. The patient was staged as FIGO 2018 clinical IIB and underwent abdominal radical hysterectomy with bilateral salpingo-oophorectomy. Intraoperative peritoneal cytology was positive.

Macroscopically, the hysterectomy specimen showed a nodular cervical lesion (Fig. 4A), and the distribution of the histologically confirmed tumor is outlined by the yellow dashed line in Fig. 4B. Serial transverse sections revealed diffuse microscopic tumor involvement extending from the cervix into the uterine corpus, vaginal wall, and parametrium (Fig. 4C). The histological tumor size was 65 mm, with full-thickness cervical stromal invasion (19/19 mm), lymphovascular space invasion, and pelvic lymph node metastasis. HE staining showed infiltrative irregular glands (Fig. 4D), and mainly PAS-positive intracellular mucin was observed with AB/PAS staining (Fig. 4E). Tumor glands were diffusely positive for CLDN18 and negative for p16 (Fig. 4F). The final pathological diagnosis was HPV-independent adenocarcinoma, GAS, pT2bN1M0 (UICC 2021), pathological stage IIIC1 (FIGO 2018). The patient received adjuvant concurrent chemoradiation with whole-pelvic irradiation and platinum-based chemotherapy and has remained free of disease for five years.

Fig. 4The alternative text for this image may have been generated using AI.

Gross and histopathological features of gastric-type cervical adenocarcinoma (GAS) in Case 2. (A, B) Gross photograph of the hysterectomy specimen opened by an anterior longitudinal incision through the cervix and vagina. (A) A nodular lesion is present in the cervix, whereas the endometrial surface appears smooth without an obvious exophytic mass. (B) The yellow dashed line indicates the distribution of the histologically confirmed tumor: GAS diffusely involves the cervix and extends into the uterine corpus, vaginal wall, and parametrium. (C) Twelve serial transverse sections of the resected uterus; the yellow dashed line denotes the distribution of microscopic tumor involvement. (D–F) Representative microscopic findings of GAS in the cervix. (D) Hematoxylin and eosin (HE) staining (200×) shows the infiltrative proliferation of irregular glands lined by mucinous columnar cells with an abundant pale-to-clear cytoplasm and variable nuclear atypia, consistent with GAS. (E) Alcian blue/Periodic acid–Schiff (AB/PAS) staining (200×) shows mainly PAS-positive (neutral) intracellular mucin with weak Alcian blue staining, highlighted as magenta/red. (F) Immunohistochemistry for p16 (200×) was negative (non–block-type) in tumor glands. (G–I) The endometrium infiltrated by GAS glands. (G) HE staining (40×) shows GAS glands infiltrating among normal endometrial glands and stroma; the black boxed area is shown at a higher magnification in (H) and (I). (H) HE staining (100×) and (I) claudin 18 (CLDN18) immunohistochemistry (100×) of serial sections show CLDN18-positive GAS glands (black arrows) interspersed between CLDN18-negative normal endometrial glands (red arrows) without an associated stromal reaction

Similar to Case 1, the morphology of the background endometrium was consistent with the proliferative phase on HE staining (Fig. 4G–H), supporting “symbiotic” invasion being the most evident in the functional layer. Microscopically, CLDN18-positive GAS glands infiltrated from the myometrium into the endometrium and intermingled with CLDN18-negative normal endometrial glands without disrupting the endometrial structure or any stromal reaction (Fig. 4I). Using the same hotspot-based approach as in Case 1 (15 fields/region at 100×; Fig. 5A–D), CD8-positive TILs density (cells/mm2) was lower in the endometrial “symbiotic” invasion area (median 110 [IQR 59.5–121.5]) than in the normal endometrium (median 314 [224.5–329.5]), the tumor center of the primary cervical lesion (median 286 [228.0–363.0]), and the invasive margin of the primary cervical lesion (median 219 [198.0–263.5]) (all P < 0.001; the Mann–Whitney U test with the Bonferroni correction; Fig. 5E). PD-1 immunostaining showed no PD-1–positive immune cells, and PD-L1 expression assessed by CPS was 0 in all four regions (Fig. 6A–H).

Fig. 5The alternative text for this image may have been generated using AI.

Comparison of CD8-positive tumor-infiltrating lymphocytes (TILs) across four regions in Case 2. (A–D) Representative CD8 immunohistochemistry (100×) illustrating the distribution of CD8-positive TILs in (A) the normal endometrium (density, 161 cells/mm2), (B) the “symbiotic” area (117 cells/mm2), (C) the tumor center (282 cells/mm2), and (D) the invasive margin (232 cells/mm2). (E) Box-and-whisker plots comparing the density of CD8-positive TILs quantified in 15 hotspot fields per region. Boxes indicate the interquartile range, with the median shown as a horizontal line; whiskers extend to the most extreme values within 1.5× the interquartile range, and dots represent outliers beyond this range. Pairwise comparisons were performed using the Mann–Whitney U test, with the Bonferroni correction for multiple comparisons (Bonferroni-adjusted significance threshold, P < 0.0083 [0.05/6]). The “symbiotic” area demonstrated a significantly lower density of CD8-positive TILs than the other regions (all P < 0.001)

Fig. 6The alternative text for this image may have been generated using AI.

Immunohistochemical staining of programmed cell death protein 1 (PD-1) and programmed death-ligand 1 (PD-L1) across different regions in Case 2. All panels are shown at 200×. (A, B) The normal endometrium stained for (A) PD-1 and (B) PD-L1; combined positive score (CPS) = 0. (C, D) The “symbiotic” area stained for (C) PD-1 and (D) PD-L1; CPS = 0. (E, F) The tumor center stained for (E) PD-1 and (F) PD-L1; CPS = 0. (G, H) The invasive margin stained for (G) PD-1 and (H) PD-L1; CPS = 0