Single Center Outcomes from Parenchymal-sparing Resections and Microwave Ablations for Neuroendocrine Tumor Liver Metastases

Background: Surgical debulking of neuroendocrine tumor (NET) is used as a therapeutic approach for metastatic NETs in selected centers. Reported outcomes after parenchymal-sparing liver resections (PSR) in NET patients with high numbers of liver metastases are sparse. Methods: NET patients that underwent surgical debulking from 2019 to 2021 were reviewed. Trends in perioperative liver function was examined, as well as symptom response, complications, and progression free survival. Results: 1069 liver lesions (median=17) were debulked from 53 patients with a combination of PSR (45%) and ultrasound-guided microwave ablations (MWA) (55%). Post-operative transaminitis was proportional to the number of lesions debulked: Median POD1 AST was 681 IU/L for 1-15 lesions vs. 1396 IU/L for >15 lesions, p=0.01 (R 2 =0.271, p<0.001) and ALT was 596 IU/L vs 1149 IU/L, respectively, p=0.01 (R 2 =0.221, p<0.001). Thrombocytopenia occurred in 75% of patients and severity correlated with increasing number of lesions (median POD2 platelets 157 x 10 9 /L for 1-15 lesions vs. 109 x 10 9 /L for >15 lesions, p=0.04; R 2 =0.163, p=0.003). Synthetic liver function measured by postoperative INR (median POD1 INR 1.3 vs 1.4, p=0.21) and total bilirubin (median POD 2 TB 1.35 vs 0.95 mg/dL; p=0.67) did not differ according to number of lesions debulked. 13% of patients sustained a Clavien-Dindo grade 3/4 complication which was not associated with the number of lesions targeted. All patients with preoperative symptoms had improvement after surgery. Median time to recurrence was 10.9 months. Conclusions:


Introduction
While generally considered rare, neuroendocrine tumors (NETs) have a rising incidence in the United States, affecting over 12,000 patients each year (1).Although described as indolent, these neoplasms exhibit signi cant metastatic potential most commonly to the liver (2).Neuroendocrine tumor liver metastases (NETLM) affect up to two-thirds of patients at initial diagnosis and progression of disease is common (3).Increasing hepatic tumor burden can lead to liver failure and death (4).Thus, control of liver disease is an important prognostic marker for these patients (5).
Available treatment options for NETLMs include systemic radiolabeled-, targeted-and cytotoxic chemotherapy, non-surgical liver directed therapy (LDT), and surgical resection.Although rarely curative, surgical debulking reduces tumor load in the liver, palliates symptoms and can dramatically improve overall survival with up to 90% 5-year survival reported (6-8).Moreover, debulking thresholds are typically set above 70% in patients with operable disease to achieve maximal bene ts (7,9).Surgical approaches to hepatic tumor debulking for NETLMs range from formal anatomic resections to parenchymal-sparing resections (PSR) which include enucleations, wedge resections and partial lobectomies (10,11).As debulking is applied to larger patient groups with increasing number of NETLMs, concerns about associated complications emerge.Reported perioperative outcomes of surgical debulking in NETLM patients have been lacking with respect to impact on liver function and morbidity proportionality to tumor burden.In this study, we report our single institutional experience in surgical debulking of NETLM patients using PSR and microwave ablative (MWA) techniques.Our objectives were to assess 1) feasibility of 90% tumor debulking in patients with large numbers of lesions, 2) safety of advanced debulking by comparing the incidence and degree of complications among patients with various numbers of lesions, and 3) the impact of debulking on postoperative liver function, symptom control and progression-free survival (PFS).

Methods
In accordance with an IRB-approved protocol, a retrospective review of NETLM patients that underwent surgical debulking at our institution from January 2019 to December 2021 was performed.All patients were under the care of our multidisciplinary NET program and were operated on by the same primary surgeon.Preoperatively, hepatic tumor burden and extrahepatic disease were assessed by abdominopelvic MRI with hepatobiliary contrast (EOVIST) and DOTATE-PET/CT ( 68 Ga and 64 Cu) imaging.A debulking goal of at least 90% was attempted when possible, and all patients had ≤ 25% liver tumor involvement, as de ned by semi-quantitative measurements (12).Operations included rst time surgical debulking with or without primary tumor resection, and re-operative debulking in patients with recurrent liver metastases after initial primary resection or prior debulking.Operations were performed using PSR, as described previously, combined with intraoperative ultrasound-guided 3-dimensional navigation MWA (Medtronic).The number of lesions targeted were delineated in operative reports.Wedge resections and enucleations were classi ed as the same technique for practicality.Percent of tumor debulked was estimated by the surgical team as the number of lesions targeted intraoperatively divided by the total number of lesions identi ed on pre-operative MRI.All patients underwent serial postoperative imaging every 3 months with MRI with EOVIST.
Various clinicopathologic data were reviewed.Since most patients had either a pancreatic NET (PNET) or small bowel NET (SBNET), comparative analyses were performed between these two groups.Speci cally, trends between laboratory values for platelets, hepatic enzymes, phosphorus, and coagulation factors in the preoperative and immediate post-operative periods were analyzed.Serum and urine tumor markers for chromogranin A and serotonin were also compared pre-and post-operatively to assess for a biochemical response to surgery which was de ned as > 50% decrease from a baseline elevated level.Post-operative symptomatic response was a secondary endpoint de ned upon review of standardized clinical documentation during follow up visits.Somatostatin analogues (SSA) were continued post-operatively for patients who remained symptomatic, had limited debulking, or G2 PNETs.Patients with progressive disease after surgery were offered subsequent therapies in the form of chemotherapy (capecitabine-temozolomide; CAPTEM), peptide receptor radionuclide therapy (PRRT), and LDT.Recurrence or progression of disease after debulking was de ned as new or evolving lesions according to the Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 standards on serial surveillance MRI (Fig. 1) (13).PFS was assessed using the Kaplan-Meier method.A univariate analysis for clinicopathologic variables was performed using Cox's proportional hazards regression.Patient characteristics, complication rates and perioperative laboratory values between patient groups were compared using linear regression, Fisher's exact, Kruskal-Wallis, and Wilcoxon Signed Rank tests.All statistical analyses were performed in R (v 4.1.1,Vienna, Austria).Perioperative morbidity and mortality were assessed using the Clavien-Dindo classi cation.

Results
Preoperative clinicopathologic characteristics of 53 NETLM patients who underwent debulking are displayed in Table 1.The majority had a well differentiated, metastatic SBNET (n = 34) or a PNET (n = 15) with grade 1 or 2 disease.One patient had a well-differentiated grade 3 NET and was one of two patients with unknown etiology of primary tumor.The remaining two patients had either a gastric or lung NET.The median age at time of operation was 57 years.43% of the patient cohort had symptomatic disease, of which 40% had carcinoid syndrome.Over 95% of patients had bilobar liver metastases, and 23% of patients had extrahepatic disease (EHD), with common sites including the peritoneum (diaphragm, omentum and pelvis) or bone.Preoperative systemic therapy with SSA, CAPTEM and/or PRRT had been given in 79%, 11%, and 7.5% of patients, respectively.With respect to liver function, a surge in liver enzymes measured on post-operative day (POD) 1 was proportional to the numbers of debulked lesions (Fig. 2A, B, C).For patients with 1-15 lesions targeted, the median POD 1 AST and ALT values were 681 and 596 IU/L, respectively, compared to 1396 and 1149 IU/L, respectively, for patients with greater than 15 lesions (Fig. 2A; p = 0.01, p = 0.01).The degree of transaminitis linearly correlated with increasing numbers of hepatic lesions debulked (Fig. 2C; p < 0.0001, p < 0.0001) but this effect typically resolved by POD 5. Synthetic liver function, assessed by changes in post-operative INR, did not differ signi cantly among patients and did not have a linear correlation with increasing number of lesions (Fig. 2C; p = 0.151).Hypophosphatemia, a common post-operative nding after liver resection, was noted to be most prominent on POD 2, but the degree did not differ among the various numbers of lesions debulked (Fig. 2C).
The median length of inpatient stay after debulking was 7 days (Table 3).Thrombocytopenia (platelets < 150 x 10 9 /L) was the most common complication occurring in over 75% of patients and most cases were transient.Two patients developed heparin-induced thrombocytopenia (HIT) and were excluded from further analyses.The severity of thrombocytopenia was proportionate to the number of lesions debulked (Fig. 2C; p = 0.003) with median POD2 platelet values 157 x 10 9 /L for 1-15 lesions vs. 109 x 10 9 /L for > 15 lesions (Fig. 2A; p = 0.04).However, the incidence of thrombocytopenia did not signi cantly differ between groups.Post-operative blood loss anemia requiring transfusion was the second most common complication, affecting 36% of patients.Active post-operative hemorrhage occurred in 4 patients as UGI bleeding (n = 2) or intraabdominal bleeding (n = 2) requiring re-exploration.Bile leaks occurred in four patients with a leak-toresected lesion ratio of 0.9%.Two cases resolved spontaneously without added intervention, one case resulted in a biloma requiring percutaneous drainage, and one case occurred after re-exploration for delayed intraabdominal bleeding and required biliary stenting.GI complications including persistent ileus (n = 2), delayed gastric emptying (n = 1) or post-operative bowel obstruction (n = 1) occurred in 4 patients.Of these, the obstructed patient required re-exploration to rule out bowel ischemia.There was one case of portal vein thrombosis that occurred in the setting of HIT and required anticoagulation.Infectious complications including urinary tract infection (n = 2), pancreatic stula (n = 3), or intra-abdominal abscess (n = 3) requiring intervention occurred in 8 patients.Overall, 13% of patients had a Clavien-Dindo grade 3 or 4 complication.While there was no 30-day mortality in our patient group, one patient died on POD 38 from NSAID-induced UGI bleeding that was managed outside of our institution.
Post-operatively, median long-term follow-up was 12.1 months (Table 4).All symptomatic patients reported symptom improvement or relief after surgery.Over 80% of patients had a biochemical response with a > 50% decrease in serotonin levels after surgery.50% of patients did not require subsequent systemic therapy at their most-recent follow-up and either had no evidence of disease (n = 17) or stable disease without progression (n = 9).The median time to recurrence after debulking was 10.9 months for all patients.There was no signi cant difference in PFS between PNET and SBNET patients (Fig. 3A 11.9 vs. 22.3 months; p = 0.095) or between patients with higher vs. lower number of lesions debulked (Fig. 3B 1-15 lesions: 22.8 months vs. >15 lesions: 16 months; p = 0.38).In cases of disease progression, ve patients required the addition of PRRT: 40% had decrease in hepatic lesion size while 60% halted in disease progression.One PNET patient required CAPTEM as subsequent systemic therapy followed by pembrolizumab for tumor de-differentiation.Seven patients underwent post-operative LDT for disease recurrence with either bland embolization or percutaneous ablation.Hazard regression univariate analysis (Table 5) indicated that among our cohort, primary tumor size, higher numbers of hepatic lesions, EHD, and other pathologic variables did not correlate with disease progression.

Discussion
Surgical debulking has been shown to palliate symptoms and prolong survival in NETLM patients (8, 14-16).Nonetheless, recurrence after debulking is universal, and both the preservation of normal liver parenchyma and minimization of complications should be prioritized for this reason.Herein, we show that PSR and MWA allows for effective debulking of hepatic disease in patients with high numbers of lesions without compromising synthetic liver function, and that such procedures can be with low morbidity in experienced centers.
Maxwell et al reported on 108 PNET and SBNET patients with NETLM who underwent 70% debulking with a median number of 6 lesions in their cohort.In their series they correlated 10 hepatic lesions with roughly a 10-19% hepatic replacement and concluded that patients with < 25% liver tumor involvement had better outcomes than those with greater tumor burden (10).Subsequently, Scott et.al demonstrated the e cacy and safety of debulking large numbers (> 10) of liver lesions using PSR and ablative techniques (17).While this study reported survival outcomes and postoperative complications across a wide range of debulking thresholds (mean 79%) with varying degrees of liver tumor involvement (5-70%) and numbers of lesions (median number 10), little is known about the effects on liver function and outcomes associated with higher degrees of surgical debulking.To address this, we analyzed our institutional data among patients with high volume of operable liver metastases who underwent 90% debulking.While the overall liver tumor involvement was less than 25%, more than half of our cohort had 16 lesions or greater.
On assessing the impact of tumor debulking on liver function, we noted that while the degree of post-operative transaminitis, indicating hepatocellular injury, correlates with the extent of debulking, synthetic liver function was not affected by higher numbers of targeted lesions.Thrombocytopenia and anemia were the most common complications after surgical debulking.Thrombocytopenia, a post-operative phenomenon that can occur after liver resection, is thought to be due to a consumptive process in liver regeneration (18-20).In our series, we noted thrombocytopenia occurred commonly, but the severity was proportional to the numbers of lesions debulked.Overall, thrombocytopenia was generally transient with nadirs reached by POD 2, and a quicker recovery was noted than what has been reported after formal anatomic liver resections (21).Post-operative anemia requiring blood transfusion occurred in over a third of patients, although active hemorrhage was rare.
Major complications de ned as Clavien-Dindo grade 3 or 4 occurred in 13% of patients, but there were no differences in type or incidence based on the number of lesions debulked.
Given our short study interval with median follow-up of  (10,24), we de ned time to recurrence according to RECIST 1.1 criteria and had comparable results with a study who also applied this model (9).For surveillance, we used MRI abdomen with EOVIST, the most sensitive method for detection of new liver lesions or progression (25) which could impact our time to recurrence.Moreover, we only initiated SSA therapy immediately postoperatively in 20% of patients.Although data remains sparse, patients on SSA after debulking surgery may perhaps bene t with extended PFS.
Various studies have reported poorer survival outcomes in NETLM patients with EHD and have questioned whether surgical debulking is bene cial (26, 27).Interestingly, the presence of EHD in our patient cohort did not correlate with disease progression.In fact, we noted that of our twelve patients with EHD, only one patient with bone metastases had signi cant disease progression after surgical debulking that required several additional therapies.Contrastingly, nine patients that had EHD limited to the abdomen and pelvis either had no recurrence or stable disease without signi cant progression at most recent follow-up.Whether or not the location and extent of EHD (i.e, bone vs intraperitoneal) impacts survival differently requires further investigation.
Among patients who had symptoms prior to debulking, all reported either improvement or relief of symptoms.At most recent follow-up after debulking, nearly half of patients did not have any disease progression.While SSA was eventually initiated in most patients postoperatively, only a few required additional treatments such as PRRT or chemo for disease progression.
Our study is limited by its retrospective nature, small cohort and short term follow up interval.Furthermore, due to the highly-specialized, multi-disciplinary and standardized care that we provide for NET patients as one of few high-volume NET programs nationwide, our outcomes may not be entirely generalizable.

Conclusions
Tumor debulking in NETLM patients with larger numbers of lesions is feasible and safe with the use of PSR and MWA.Perioperative outcomes such as complications, mortality, and synthetic liver function are not affected by higher numbers of lesions and most patients have symptom relief after debulking.Thus, this therapy should be considered for NETLM patients with larger number of hepatic lesions as it may offer symptom and disease control.

Table 1 .
"Comparison of NET patients preoperative clinicopathologic characteristics" Intraoperative characteristics are reported in Table2.A total of 1069 hepatic lesions were debulked collectively with PSR (45%) and MWA (55%).Strati ed by number of lesions, 21% (n = 11) had up to 5 lesions, 23% (n = 12) had 6-15, 28% (n = 15) had 16-25, and 28% (n = 15) had > 25 lesions debulked.The highest number of lesions debulked in one patient was fty-six.The median number of lesions targeted was 17 overall.All patients, except for two, had > 90% tumor debulking at the time of operation.Over 70% of patients had synchronous resection of the primary tumor as their rst operation, which they had on average 6.5 months after initial diagnosis.The average estimated blood loss (EBL) was 300 ml for all cases, 13% of which required at least 1 intraoperative transfusion of packed red blood cells.Table 2. "Operative characteristics for NETLM patients by tumor type"

Table 3 .
"Morbidity and Mortality for patients per NET type and number of lesions debulked".LOS = length of stay; EBL = intraoperative estimated blood loss

Table 4 .
"Postoperative characteristics after tumor debulking by NET type".
rd adjuvant therapy

Table 5 .
"Univariate analysis of clinicopathologic variables associated with progression free survival." 12 months, we assessed PFS rather than overall survival (OS).Although SBNETs seemed to have a longer PFS than PNETs, a trend that other studies have reported (1, 10, 22, 23), our results did not reach statistical signi cance and may be due to multiple reasons, including smaller earlier detection of PFS due to better imaging, and perhaps even the difference in routine usage of SSA postoperatively.The median time to recurrence overall was 10.9 months, and while this is lower than what has been previously reported