Microangiopathic Hemolytic Anemia Is a Late and Fatal Complication of Gastric Signet Ring Cell Carcinoma: A Systematic Review and Case-Control Study

Abstract Background Microangiopathic hemolytic anemia (MAHA) is a rare paraneoplastic syndrome that has been reported in patients with gastric signet ring cell carcinoma (SRCC). Clinical and prognostic features of MAHA in this setting have been poorly described. Materials and Methods We conducted a systematic review in 8 databases of gastric SRCC complicated by MAHA and performed a case-control study assessing factors associated with survival in patients with gastric SRCC and MAHA in our pooled cohort compared with age-, sex-, and stage-matched cases of gastric SRCC from the Surveillance, Epidemiology, and End Results (SEER) database. Descriptive analyses were performed and multivariable Cox-proportional hazards regression modeling was used to determine factors associated with overall survival. Results All identified patients (n = 47) were symptomatic at index presentation, commonly with back/bone pain, and dyspnea. Microangiopathic hemolytic anemia was the first manifestation of gastric SRCC in 94% of patients. Laboratory studies were notable for anemia (median 7.7 g/dL), thrombocytopenia (median 45.5 × 103/μL), and hyperbilirubinemia (median 2.3 mg/dL). All patients with MAHA had metastatic disease at presentation, most often to the bone, bone marrow, and lymph nodes. Median survival in patients with gastric SRCC and MAHA was significantly shorter than a matched SEER-derived cohort with metastatic gastric SRCC (7 weeks vs 28 weeks, P < .01). In multivariate analysis, patients with MAHA were at significantly increased risk of mortality (HR 3.28, 95% CI 2.11-5.12). Conclusion Microangiopathic hemolytic anemia is a rare, late-stage complication of metastatic gastric SRCC and is associated with significantly decreased survival compared with metastatic gastric SRCC alone.


Introduction
Gastric cancer is the third leading cause of cancer-related death worldwide. 1 Signet-ring cell carcinoma (SRCC) is a rare subtype that accounts for ~15% of gastric adenocarcinoma cases and is characterized histologically by the presence of distinct tumor cells with mucinous cytoplasm and crescent-shaped nucleus. [2][3][4] Signet-ring cell carcinoma tends to have an aggressive course and commonly presents at an advanced stage. 5,6 Microangiopathic hemolytic anemia (MAHA) is a rare paraneoplastic syndrome described in patients with solid tumors, including metastatic SRCC. Tumor cell emboli, immune complexes, and tumor-derived factors are thought to contribute to the Coombs-negative hemolytic process that occurs. 7 The diagnosis is made when schistocytes are identified on peripheral blood smear. 8 The differential diagnosis for MAHA includes disseminated intravascular coagulation, thrombotic thrombocytopenic purpura, hemolytic uremic syndrome, heparin-induced thrombocytopenia, malignant hypertension, HELLP (hemolysis, elevated liver enzymes, and low platelets) syndrome, antiphospholipid syndrome, and drug-induced thrombotic microangiopathy. 9 Familiarity with the clinical features of this rare complication including laboratory testing, radiologic findings, and treatment response may lead to prompt and accurate diagnosis and prognostication. To this end, we report a case of metastatic gastric SRCC complicated by MAHA according to CARE guidelines, 10 perform a systematic review and synthesis of individual patient data of reported cases, and compare gastric SRCC complicated by MAHA to age-, sex-, and stage-matched gastric SRCC in general via regression analysis incorporating data from the Surveillance, Epidemiology, and End Results (SEER) database.

Case Presentation
A 36-year-old previously healthy woman was admitted to a major academic medical center after a presyncopal episode associated with nausea, vomiting, and diaphoresis. She also endorsed a 1-week history of new lower back pain. She denied any signs of gastrointestinal bleeding. Initial vital signs were temperature 100.5°F, blood pressure 107/45, respiratory rate 16/minute, and SaO 2 of 100% on room air. Physical examination was notable for tenderness in the lumbar area with no neurologic abnormalities.
Although lumbar radiographs were normal, computerized tomography (CT) scan of the chest, abdomen, and pelvis identified supraclavicular lymphadenopathy and lytic lesions of the L1 vertebral body with endplate compression fractures. Magnetic resonance imaging (MRI) of the spine, abdomen, and pelvis revealed multifocal enhancing osseous lesions throughout the spine consistent with metastatic cancer. Tumor markers CEA and CA19-9 were elevated to 16.3 ng/mL and 4819 U/mL, respectively. Image-guided bone biopsy of the L1 lytic lesion was notable for bone marrow infiltrated by poorly differentiated carcinoma cells that had intracytoplasmic mucin vacuoles and ill-formed glandular structures (Fig. 1A). The tumors cells were positive for AE1/AE3, CK7, CK20, and CDX2 immunostains, but negative for GATA3, PAX8, CD20, and CD3; this was consistent with metastatic carcinoma of gastrointestinal origin (Fig. 1B, C). A bone marrow biopsy showed marked hypocellularity and highly necrotic tumor positive for AE1/AE3, similar to the bone biopsy lesion with metastatic carcinoma (Supplementary Fig. S1A-C). The tumor was negative for Her2 and PDL-1.
Given the bone biopsy findings, upper endoscopy and colonoscopy was performed which showed a non-bleeding ulcer in the gastric body ( Supplementary Fig. S2) with a normal colon. Biopsies from the gastric ulcer showed a poorly differentiated adenocarcinoma with signet ring cell features (Fig. 2). Primary breast cancer was ruled out with a normal BIRADS-1 (Breast Imaging Reporting and Data System) breast ultrasound and mammogram. The patient was diagnosed with MAHA secondary to gastric SRCC. Treatment of spinal metastases was initiated with dexamethasone and palliative radiation to the cervical and lumbar spine. During her index hospitalization, she received one cycle of dose-reduced FOLFOX (fluorouracil , leucovorin, oxaliplatin) with prolonged hospitalization complicated by bilateral spontaneous subdural hematomas with midline shift and altered mental status. She was ultimately discharged after 30 days of hospitalization and completed an second cycle of FOLFOX as an outpatient.
Unfortunately, despite cancer-directed treatment and aggressive supportive care, the patient became transfusion-dependent, and her bony pain became increasingly severe, requiring rehospitalization. Interval radiographs of the patient's pelvis and femur confirmed extensive metastatic disease, and MRI of the lumbar spine demonstrated new pathologic L4 vertebral body compression fractures. Given the patient's widespread metastatic disease, evidence of progression, poor quality of life, and grim prognosis, she was transitioned to comfort measures and died 10 weeks after her initial diagnosis.

Systematic Review
The systematic review is reported according to the guidance of the preferred reporting items for systematic reviews and meta-analysis (PRISMA) statement and statement for reporting and synthesis without meta-analysis with an a priori protocol (Supplementary Tables S1-S3). 11,12 The study was registered in PROSPERO as study CRD42021240057. Considering the rarity of this paraneoplastic phenomenon, we used a mixed methods design that combines health systems data with a systematic review of the literature to identify the largest number of unique cases as recommended by the methods guide of the Agency for Healthcare Research and Quality. 13

Search Strategy and Data Sources
A systematic search was conducted by a medical librarian (A.G.) in Cochrane Library, EMBASE, Google Scholar, Japan Medical Abstracts Society, Ovid MEDLINE, PubMed, Scopus, and Web of Science Core Collection databases using various structured terms and text words without language restriction for studies reporting cases of signet cell carcinoma and MAHA on March 2, 2021. Reference lists of relevant papers were also screened manually for additional cases. Details regarding the search strategy are provided in Supplementary Table S4. Inclusion criteria included adult patients aged 18 or older with biopsy-confirmed gastric SRCC and laboratory evidence of MAHA. We excluded duplicate cases or cases in which there was insufficient reporting of clinical data of individual patients (Supplementary Table S5). Japanese studies were extracted by a native Japanese clinician (M.H.). Other non-English studies were translated with Google Translate (https:// translate.google.com) with assistance from a native language speaker, if needed.
The methodological quality and synthesis of case series and case reports tool was used to evaluate included reports (Supplementary Table S6), and this tool has been applied previously with consistency among reviewers. 14-16 Study selection, data extraction, and risk of bias assessment were made by 2 independent reviewers (R.L., N.T.), and disagreements were settled by discussion and adjudication by the corresponding author (D.K.L.). Study selection was conducted using Covidence (https://www.covidence.org/), and data extraction was completed in REDCap (https://www.project-redcap. org/). The data extracted included year of publication, publication format, country of origin, publication language, age, sex, ethnicity, medical history, clinical symptoms at presentation, laboratory findings, endoscopic findings, radiological exams, treatment, post-treatment complications, duration of follow-up after therapy, final outcome, and cause of death.

Surveillance, Epidemiology, and End Results Database Patient Selection
The SEER database is an open public database launched by the National Cancer Institute in 1973. It is the authoritative source of clinical information regarding cancer patients in the US. We used the SEER 18 database, which includes a full set

754
The Oncologist, 2022, Vol. 27, No. 9 of 18 registries and has the broadest coverage of clinical data between years 2000 and 2018.
Eligible controls selected from the SEER database had to be adults aged ≥19 with a diagnosis of distant signet ring gastric cancer (ICD-O3: histology 8490/3 and site 16.x) between 2000-2017. We excluded patients who: (1) were diagnosed with cancer-based on autopsy or death certificate, (2) were alive with no survival time data, and (3) another cancer diagnosis. A total of 6607 newly diagnosed distant SRCC of gastric origin were identified as potential controls in the final SEER cohort. SEER*Stat 8.3.9.2 was used to identify patients.

Variables
Among eligible patients, we extracted data for the following covariates: sex, race, age at diagnosis, year of diagnosis, survival time, survival status, and treatment data (no treatment, chemotherapy only, other treatment). The outcome of interest was overall survival, as defined from diagnosis to death or last follow-up visit.

Matching Process
Our pooled cohort of patients was matched to controls from the SEER database in a 1:4 ratio based on sex, age, and year of diagnosis. Up to 4 years of variation between cases and matched controls were permitted to minimize era-specific differences in management.

Statistical Analysis
For descriptive analysis, we reported medians and ranges for continuous variables and percentages for dichotomized variables unless otherwise specified. A comparison of outcomes was performed using Pearson's χ 2 test. When the number of patients in any cell in a contingency table was under 5, we used Fisher's exact test. Two-tailed P-values were statistically significant when they were below 0.05. Multivariable Cox regression models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for risk factors related to overall survival. Kaplan-Meier analysis was used to estimate cumulative survival probability. All analyses were performed using SAS (Version 9.4, SAS Institute, Cary, NC).

Study Characteristics
A flow diagram of the study selection in shown in Fig. 3. A total of 26 publications between 1985 to 2021 were identified that met the study selection criteria. Eighteen publications were in English, 5 in Japanese, 2 in German, and one in Spanish. Details regarding the countries of origin for reach publication can be seen in Supplementary Table S7.

Assessment of Methodological Quality of Included Studies
Reviewer agreement in the methodological quality of included studies was 100%. Most studies showed an unclear risk for selection bias. There was low risk of ascertainment, causality, and reporting bias (Supplementary Table S8).

Demographics and Baseline Laboratory Studies
We identified 47 patients with SRCC of gastric origin and MAHA (Table 1) were consistent with a consumptive, hemolytic process, specifically low hemoglobin (median 7.7 g/dL, IQR 2.2 g/dL), low platelets (median 45.5 × 10 3 /μL, IQR 28.0x10 3 /μL), elevated total bilirubin (median 2.3 mg/dL, IQR 2.6 mg/ dL), low haptoglobin (median 0 g/L, IQR 0.01 g/L) and high lactate dehydrogenase (median 774 IU/L, IQR 1184 IU/L). In 7 cases, carcinoembryonic antigen (CEA) was checked and was found to be markedly elevated to median 23.4 ng/ mL (IQR 63.25 ng/mL). In 29 patients for which the type of MAHA was further subclassified, MAHA was classified as disseminated intravascular coagulation (DIC) and thrombotic thrombocytopenic purpura (TTP) in 86.2% and 13.8% of patients, respectively. A detailed description of baseline laboratory values, treatment and outcome data for each patient is provided in Supplementary Table S9.

Treatment and Response
Information regarding treatment was available for 41 patients and are summarized in Table 2. Notably, 46.3% of patients did not receive any cancer-directed treatment. Cytotoxic chemotherapy (41.5%) was the most common in those who received treatment. Specific chemotherapeutic therapies that were used included 5-FU-based chemotherapy (94.4%) and cisplatin-based chemotherapy (72.2%). Details of the chemotherapy regimens used in our pooled cohort are provided in Supplementary Table S12. Most patients (57.1%) did not have any clinical or radiologic response despite treatment, the latter defined as reduction of tumor burden based on follow-up imaging.
In terms of MAHA response, only 60% (9/15) of patients improved with therapy defined by amelioration in presenting anemia and/or thrombocytopenia. Notably, 78% (7/9) of patients remained transfusion-dependent even after treatment.

Follow-up and Outcomes
Information with regards to survival outcomes was available for 45 patients. The median follow-up time was 8 weeks (range 0.1 to 76 weeks) after the presentation. Overall mortality was 91%. The overall cause of death was most commonly due to hemorrhage (61.1%), respiratory failure (22.2%), and carcinomatosis (11.1%).

Survival Analysis
Next, we sought to understand the impact of MAHA on overall survival time by comparing our pooled cohort to patients with gastric SRCC with unknown MAHA status in a case-control analysis. From the publicly available SEER database, we identified a total of 6607 potential control patients with metastatic gastric SRCC with unknown MAHA status. Selection of our pooled cohort for the analysis excluded cases diagnosed before the year 2000 (11 cases) and those with missing Given that all cases from the pooled cohort had metastatic disease, and control cases were selected from a cohort of metastatic gastric SRCC, these 2 groups were effectively matched for stage. Regression analysis was performed to identify characteristics that influenced survival. Covariates chosen from individual patient data from our systematic review and meta-data from SEER included age, sex, ethnicity, year of diagnosis, and follow-up time.
Baseline characteristics of the pooled cohort and SEERderived control cohort are shown in Table 3. There were no significant differences in baseline parameters for both groups. After multivariate regression, we found that the presence of MAHA (hazard ratio [HR] 3.28, P < .01) and receiving no treatment (HR 3.57, P < .01) were associated with a significantly increased risk of mortality.
Kaplan-Meier survival curves for metastatic gastric SRCC and metastatic gastric SRCC complicated by MAHA is shown in Fig. 4. Survival in the MAHA cohort was significantly less than that of the SEER cohort (7 vs 28 weeks; P < .01).

Discussion
Our study highlights MAHA as a rare, late-stage complication of gastric SRCC. MAHA is often the first overt manifestation  of metastatic SRCC. In fact, all patients in our analysis had metastatic disease at index evaluation, commonly to the bone, and lymph nodes. Endoscopy, bone marrow biopsy, and CT imaging are useful diagnostic studies to identify the origin of the primary cancer. Even with treatment, the overall prognosis is poor, with higher mortality rates and worse survival than patients with metastatic gastric SRCC without MAHA. Cancer-related MAHA (CR-MAHA) is a type of Coombsnegative thrombotic microangiopathy observed in solid cancers and is characterized by erythrocyte fragmentation. Evaluation will reveal schistocytes on blood smear, elevated LDH, low or absent haptoglobin, thrombocytopenia, and elevated total bilirubin levels. 17 Schistocytes are fragments of damaged red blood cells formed from the excessive wall shear stress on red blood cells in MAHA; they can be visualized on blood smear as helmet cells or irregular, triangular, and crescent-shaped cells lacking central pallor. 18 Among CR-MAHA, gastric cancer is the most common cancer type, and adenocarcinomas are the most common histologic type. 7 The reason for this is unknown. Signet ring cell carcinoma comprises 16.8% of all gastric cancer cases and has a strong predilection for gastric tissue. 19 While the pathogenesis of MAHA and gastric SRCC remains unclear, several mechanisms are postulated to explain their association. Signet ring cell carcinoma is an adenocarcinoma that produces mucin, which has erythrocytopathic activity. 20 Moreover, mucinous tumors secrete enzymes that can activate factor X and provoke DIC. 21 In patients with solid neoplasms like gastric cancer, hemolysis primarily occurs from the mechanical shearing of red blood cells passing through small blood vessels, especially in infiltrated organs such as the bone marrow. 22,23 Direct invasion of the endothelium may also lead to cell injury, triggering thrombotic microangiopathies. 24 Finally, tumor cells can produce cytokines which activate tumor necrosis factor-mediated red cell damage. 25 Our case-control analysis using SEER data showed that patients with metastatic gastric SRCC and MAHA have worse survival than age and sex-matched control patients with metastatic gastric SRCC (MAHA status unknown). We propose that MAHA is a surrogate marker for metastatic disease, primarily responsible for serious, fatal outcomes. It is well established that patients with diffuse metastatic gastric adenocarcinoma, including the SRCC type, have a poor prognosis. [26][27][28] Of note, all our patients had metastatic disease at initial presentation. Increased risk of mortality and short survival time is common in disseminated gastric SRCC and MAHA cases. Of note, MAHA associated with cancers of other primary sites also show very similar poor outcomes. In one of the largest reviews of CR-MAHA, survival was poor in cancers originating from the breast (median 0.5 months without chemotherapy/surgery, 4 months with chemotherapy/surgery), lung (median 0.5 months without chemotherapy/surgery, 5 months with chemotherapy/surgery), and unknown primary (median 0.5 months without chemotherapy/surgery, 3.5 months with chemotherapy/surgery). 7 Plasma exchange and fresh frozen plasma was rarely effective in patients with gastric, breast, lung or unknown primary cancers associated with MAHA. Nearly all the patients in that study also had metastatic disease, further suggesting that metastatic disease as the factor responsible for the demise of affected patients.
Palliative treatments exist for both metastatic SRCC and MAHA, although simultaneous management of both conditions can be competing and challenging. Treatment for metastatic SRCC includes chemotherapy and radiation. 26 Chemotherapy should be started as soon as possible. 29 Blood products transfusion is often required. 22 Combination chemotherapy has been well-established in the palliative treatment of advanced-stage gastric cancer and includes fluoropyrimidine (5FU or capecitabine) and oxaliplatin, fluoropyrimidine (5FU or capecitabine) and cisplatin; for patients with HER2 positive tumors, adding trastuzumab is recommended and for patients with tumors that are CPS > 5, adding nivolumab is recommended. 30 In our cohort of patients, most had chemotherapy regimens consisting of 5-flurouracil and cisplatin. TEFOX (docetaxel-5FU-oxaliplatin) has been evaluated as an alternative option for SRCC in a recent retrospective study. 31 However, prognosis remains poor due to extensive metastatic disease, short remission, and complications from MAHA (like hemorrhage) which can be life-threatening. In fact, for many patients, they may already be too ill and lack the clinical stability to initiate chemotherapy.
The clinical significance of our study lies in highlighting the most common clinical phenotype of a patient with gastric SRCC and MAHA at presentation. Our study also underscores the inisiduous nature of the disease process given that MAHA was present when the initial diagnosis of gastric SRCC was made for nearly all the patients in our cohort. If a patient presents with back and bone pain along with evidence of hemolytic anemia, advanced stage gastric SRCC with MAHA should be considered on the differential and an EGD can be considered. Similarly, a finding of MAHA in should prompt the clinician to rule out gastric cancer in patients at risk. If the latter is found with MAHA, thorough cross-sectional imaging needs to be pursued given this association between MAHA and distant metastases. Additionally, the summary of our study findings can provide prognostic information to patients and families. It should be emphasized that prognosis is potentially poor and aggressive therapy may not yield significant survival benefits. A multidisciplinary discussion which involves gastroenterology, oncology, and palliative care medicine is thus warranted.
A key strength of our study is the systematic methodology used to capture all gastric SRCC and MAHA cases in the literature to describe the clinical presentation, evaluation, and outcomes of this rare phenotype. We also used the large publicly available SEER database to identify stage, age, and sex-matched controls and highlight differences in metastatic gastric SRCC survival with and without MAHA. Another strength of our study was in implelemting a novel methodology of combining a case study, systematic review, and case-control analysis approach. 16,32 A systematic review alone may be insufficient to provide guidance in informing decisions in care. 13 Combining health system data with a systematic review approach can further strengthen the evidence and enhance the applicability of systematic review findings. 13 In our study, the systematic review provided a comprehensive context to understand the patient's overall disease course and medical management in the case study. Similarly, the case study and case-control analysis were complimentary to the systematic review by expanding the evidence base given the limited number of total cases in the literature.
Our study has several inherent limitations that should be acknowledged. Several data parameters were missing and reporting bias is likely present. Importantly, selection bias is likely present given that individual patient data was derived from case reports and case series. However, evidence from case reports and series takes on increased significance when higher levels of evidence are not available. 14 In addition, given patients with gastric SRCC and MAHA have short survival time with high rates of mortality, patients may have perished prior to work-up and were thus not reported. Moreover, an era effect is likely present given the inclusion of data published between 1985 to 2020, with 2011 as the median year of publication. Since there have been advances over time regarding the diagnostic and therapeutic approaches, earlier study findings may not be generalizable for patients in the current day and age. However, our case study survival outcome was similar to that of the overall cohort even when cases diagnosed before 2000 were excluded for the case-control analysis. Lastly, it is important to note that the MAHA status within our SEER cohort was unknown. However, a significantly longer survival time of SEER cohort compared with our MAHA-proven cohort was demonstrated, despite matching of other important parameters.

Conclusion
In conclusion, MAHA is a late-stage complication of gastric SRCC that is a surrogate marker for metastatic disease. Cancer-directed treatment and treatment for MAHA both appear to provide minimal benefit. Overall prognosis is poor with short survival despite treatment.

Data Availability
The data underlying this article will be shared on reasonable request to the corresponding author.