This study evaluated and characterized ascites in subjects with primary extrahepatic solid tumors including breast cancer, lung cancer, gastric cancer, pancreatic cancer, ovarian cancer, colon cancer and renal cell cancer. The most common etiology of ascites was peritoneal carcinomatosis based on fluid analysis (62%) or cytology/imaging (79%). Similar to our data, a prior study of 45 patients reported 53% of malignancy-related ascites was due to peritoneal carcinomatosis also across a broad range of malignancies . High SAAG with high total protein accounted for 29% of subjects with malignancy and ascites. Among the patients with high SAAG suggesting noncirrhotic portal hypertension based on ascites fluid analysis, 67% had evidence of peritoneal carcinomatosis on cytology or imaging suggesting that multiple mechanisms are contributing to the development of ascites. Interestingly, 9% had low SAAG and low ascites fluid total protein levels with no evidence of nephrotic syndrome, pancreatic ascites, tuberculous peritonitis, or serositis. The median serum albumin level in this group was low 1.99 g/dl (IQR 1.8-2.03) likely as a reflection of cachexia and other factors and may explain the low SAAG findings.
Liver Metastases Versus No Liver Metastases:
In those with ascites and no liver metastases, peritoneal carcinomatosis was the most common etiology of ascites in this group at 61%, and lung cancer was the most common malignancy in this group at 27%. Significantly more subjects without liver metastases and no evidence of peritoneal carcinomatosis on cytology or imaging had peritoneal fluid reflecting portal hypertension with high SAAG, 16/119 (13%) compared to subjects with liver metastasis 7/122 (6%). A portion of these individuals (3/16 and 2/7 respectively) received chemotherapies known to be associated with non-cirrhotic portal hypertension which may in part explain these findings. Chemotherapy/immunotherapy administration alone was the most common approach to manage ascites in the no liver metastases subjects with 37% of subjects receiving this type of therapy.
Among subjects with ascites and liver metastases, peritoneal carcinomatosis was again the most common etiology of ascites in this group at 63%, a rate similar to those without liver metastases, though breast cancer was the most common malignancy in this group at 24% (Table 3). In one malignant ascites series, breast cancer was also reported to be the most common extrahepatic tumor associated with ascites . Peritoneal carcinomatosis was commonly present with liver metastases suggesting that the presence of liver metastases reflects a higher overall disease burden. Drainage of peritoneal fluid by paracentesis or indwelling catheter combined with chemotherapy/immunotherapy was the most common method of managing ascites in these subjects at 33% suggesting that fluid accumulation may be more difficult to control.
We found that the 5-year mortality rate in subjects with liver metastases was significantly higher at 61% compared to 45% in subjects without liver metastases regardless of ascites characteristics with no differences when examined in 5-year period increments (Figure 1a). The presence of liver metastases was reported to be an independent predictor for mortality in patients with solid tumors who developed tumor lysis syndrome . Similar to our findings, one prior report has noted that the presence of liver metastasis in subjects with malignant ascites was an independent prognostic factor associated with poor outcomes 
Ascites Reflecting Peritoneal Carcinomatosis Or Portal Hypertension:
Peritoneal carcinomatosis was the most common etiology of ascites in our cohort with lung cancer being the most common malignancy in subjects with ascites due to PC at 24%. We noted that patients with ovarian cancer had ascites consistent with peritoneal carcinomatosis only. To date, no specific solid tumor has been associated with a higher rate of peritoneal carcinomatosis. In the entire cohort, systemic chemotherapy or immunotherapy alone was the most common management approach of ascites in subjects with peritoneal carcinomatosis 52/150 (35%). The most common approach in those with peritoneal carcinomatosis and no liver metastases (41%) was to administer chemotherapy/immunotherapy alone, whereas the majority of subjects with peritoneal carcinomatosis and liver metastases (31%) required paracentesis in addition chemotherapy/immunotherapy indicating multiple therapies may be required in the setting of malignancy and hepatic metastases to control the ascites. We noted a higher 1 and 3-year in those with ascites due to peritoneal carcinomatosis (Figure 1B, 1C, 1D). In Subgroup analysis examining mortality rates by tumor type, subjects with pancreatic cancer and evidence of PC on cytology/imaging had higher 5-years and 1-year mortality rates compared to subjects without PC (Figure 1E, 1F). A prior study also reported poor prognosis in subjects with pancreatic cancer and peritoneal carcinomatosis .
Low SAAG ascites has been reported to be caused by peritoneal carcinomatosis in addition to tuberculous peritonitis, nephrotic syndrome, and pancreatitis [1, 11]. We identified 20/241 (8%) subjects with low SAAG and low total protein, and all had evidence of peritoneal carcinomatosis on cytology and or imaging. Low SAAG ascites has been reported in up to 20% of those with malignant ascites in prior reports and in our cohort, the serum albumin levels were low in our cohort (median 1.99, IQR 1.83-2.13) . In our cohort, the diagnostic accuracy of high ascites fluid total protein in diagnosing PC was 75%, similar to prior studies .
Interestingly, evidence of portal hypertension was present in 33% of subjects with no liver metastases and 24% of subjects with liver metastases with pancreatic cancer being the most common malignancy in subjects with ascites due to portal hypertension at 23 %. Significantly more subjects with renal cell cancer had high SAAG ascites (13/69; 19%) compared to (11/150; 7%) in the PC group with just 3/13 subjects receiving azathioprine, oxaliplatin, trastuzumab, or emtansine, all therapies associated with non-cirrhotic portal hypertension. In our retrospective series, chemotherapy or immunotherapy alone was the most common management approach for the ascites in subjects with evidence of portal hypertension 23/69 (33.3%). No liver biopsies were performed in the subjects without liver metastases group to determine if sinusoidal infiltration of tumor, nodular regenerative hyperplasia or other etiology was present nor were hepatic venous pressure measurements performed. We noted limited use of diuretic therapies in those with high SAAG ascites (11%), though diuretics should be a first line consideration in this subgroup . There are limitations to our study. We reported a single center and retrospective data without a standardized approach. Not all diagnostic tests were done in all subjects. No liver biopsies or pressure measurements were performed to explain or confirm the findings of portal hypertension.
In summary, although ascites in solid tumor malignancy is most commonly due to peritoneal carcinomatosis, we observed characteristics of portal hypertension with SAAG>1.1 in 29% of subjects. We also observed that the presence of liver metastasis was not the sole contributor to the development of ascites with features of portal hypertension which could be related to prior chemotherapies or other factors. With ascites fluid analysis, we believe that ascites in the setting of extrahepatic malignancy can be managed with a combination of therapies directed toward the tumor, especially in the era of increasing targeted therapies, and if present portal hypertension . We also noted that despite advances in targeted chemo/immune-oncology therapies and the decreased mortality rates in cancer patients, patients with evidence of PC continues to have high mortality rates. Figure 2 reflects a proposed algorithm for the management of ascites in subjects with hepatic solid tumors. Finally, we noted multiple presentations that predicted higher mortality including those with liver metastasis or peritoneal carcinomatosis on cytology and or imaging. These findings may help inform prognosis and treatment strategies.