The most common underlying disease in our study was hepatic metastases of a primary extrahepatic tumor (54.3%). HCC was the second most common indication for SIRT (39.7%), while BTC contributed to 6% of the cases. There were significant changes within the spectrum of etiologies during the observational period regarding HCC (mean: +1.2%/year) and filiae (mean: -1.7%/year). BTC (mean: +0.5%/year) tended to increase but the differences were not statistically significant. To our knowledge, there is no published population- or geographical-based data about the distribution of SIRT in terms of different tumor entities.
Current guidelines mention SIRT in HCC patients with BCLC-0 as equally effective compared to TACE if ablation or surgical resection is not feasible in patients with solitary HCC ≤ 8 cm [14, 15, 23]. In patients with BCLC-A, SIRT could be considered in some patients as a bridge to resection or transplantation in larger tumors [23]. TACE, systemic therapy, and transplant in selected patients are the backbones for BCLC-B. SIRT can be indicated for intermediate (BCLC-B) or advanced-stage HCC patients who are poor candidates for TACE because of massive tumor size, bilobar disease, or portal vein thrombosis (PVT) [14]. The status of SIRT in patients with BCLC-C remains arguable after several studies (SAHaRA, SIRveNIB) comparing SIRT and former systemic first-line therapy (Sorafenib) failed to show survival benefit [24, 25], especially against the background of newer promising systemic therapies. It is important to mention, that comparative data for SIRT vs. TACE in the context of HCC and BTC is generally low and based on small retrospective analyses showing a benefit in PFS and QoL for HCC patients but lacking a benefit in OS [26]. According to current guidelines SIRT can be used (as well as other trans-arterial procedures, e.g. TACE) in liver-limited advanced cholangiocarcinoma if systemic therapies are not feasible or as an adjunct to systemic therapies and in locally limited intrahepatic cholangiocarcinoma in second-line therapy for selected patients [15–18]. The guideline recommendations regarding SIRT in HCC and BTC did not change essentially in recent years and the incidence of HCC in Germany did not change during the observational period [27], while incidence of intrahepatic cholangiocarcinoma (ICC) is rising, but the overall incidence for BTC remained constant [9, 18]. Of note, the increasing number of SIRTs in the context of HCC and BTC in our study reflects the current guidelines that still recommend SIRT in the context of HCC but not (or only as “last resort” treatment) in other entities. There was no disproportionate rate of PV in our cohort with only 1.7% (n = 184) which could explain the differences. However, it must be pointed out that we could not distinguish between the BCLC stages and the intention to treat (bridging/down-sizing vs. palliative setting) but the relatively high percentage of hospitals with a low number of SIRTs/year suggests the latter, at least in a relevant portion of the cohort.
Worldwide Data on the use of the different radiation agents, and especially 166Ho comparative data in different settings are missing. Holmium-based SIRTs were introduced into clinical routine much later than yttrium-based SIRT. Thus in our analysis, the vast majority of SIRT procedures were performed with ⁹⁰Y (99.6%) over 166Ho (0.4%). This finding is possibly due to the overall long-term experience of the centers in using yttrium-based SIRTs as the standard of care. The first 166Ho-SIRTs in our study were recorded in 2017 and the number of 166Ho-SIRTs increased significantly through the years 2017 to 2019, despite overall being low, accounting for 2% of all SIRTs in 2019. Worldwide data about the use of different agents and comparative data between ⁹⁰Y- and 166Ho-based SIRTs would be helpful.
Overall SIRT is a safe procedure with very low overall mortality with a well-definable spectrum of adverse events. The overall in hospital-mortality in our study was 0.14% and is, to our knowledge, the first published in hospital-mortality rate in a large cohort. The reported treatment-related mortality rates are single-center experiences and mid-to-long-term SIRT-associated complications can occur weeks after the procedure, mostly after hospital discharge, ranging from 0% − 4% [31, 32]. As pointed out in detail later, the assignment of a complication to the SIRT is not possible without doubt on the basis of our data, representing a major limitation of our study. It should be emphasized that our study did not report the cause of death or long-term adverse events as we only were able to report the in-hospital course.
Despite being a minimal-invasive local therapy with the main effect based on the short range of radiation in the tumor tissue, a small amount of radiation is also delivered to the lungs (lung shunt). The procedure may only be performed if the expected lung shunt calculated based on pretreatment planning does not exceed a certain threshold. Any other (non-pulmonary) extrahepatic, non-target deposition of microspheres are considered complications and may result in substantial radiation-induced disease of the affected non-target organs. To avoid any extrahepatic, non-target deposition thorough patient planning including (99mTc-MAA-scanning; holmium scout dose) is mandatory prior to SIRT. Non-target complications may include radiation gastritis and gastrointestinal ulcers, cholecystitis, radiation pneumonitis, and radioembolization-induced liver disease (RAILD) which may occur despite thorough pretreatment planning [33]. The overall most common complication in our study was gastrointestinal (ulcers, gastritis, duodenitis) in 2.7% of all cases. Gastrointestinal deposition of microspheres during the SIRT procedure due to the communication of liver arteries and the digestive tract via collaterals, reflux, catheter dislocation or direct injection in close digestive arteries can lead to radiation ulcers and gastritis [34, 35]. Ulcera ventriculi et duodeni occurred in 0.4% of the cases each, which is lower than the previously reported data: In the largest clinical trial on SIRT, the incidence of GI ulcers was 2.4% among patients with metastatic CRC [36]. A possible explanation for the lower rates in our study is the partially delayed onset of ulcers partially in a period of a few weeks and the comparatively short observational period in our study. Gastritis was the overall most frequent single complication with a total number of 193 cases (1.8% of all SIRTs). Duodenitis occurred in 20 cases (0.18% of all SIRTs). The overall cases are slightly higher but in line with the published data: Kennedy et. al reported a rate of 1% for gastritis [37]. More distal injections of microspheres can be performed to prevent these complications [38], but we did not record the rate of coil embolization or the exact localization of the injection.
Acute cholecystitis is a rare complication of SIRT. In an analysis of CT changes in the gallbladder following (20–30 days) SIRT, asymptomatic thickening and hyperenhancement of the gallbladder wall were observed in 10 out of 42 patients [39], but the rate of clinical apparent symptoms of acute cholecystitis is low and only about 10 cases of acute cholecystitis have been reported in the literature in detail [33, 40]. Kennedy et. al[37] reported a rate of 1.3% (8 cases). In our study, there was a slightly lower rate of 0.3%.
Acute kidney injury occurred in 0.3% and pulmonary events (pneumonia, pulmonary artery embolism) in 0.2% of all performed cases. Further complications were sepsis in 0.17%, pancreatitis in 0.1%, and liver abscess in 0.04% of all cases. Catheter-related complications (aneurysm, major bleeding) occurred in 7 cases (0.06%). To our knowledge, there are no published data about the frequency of these adverse events in a large patient cohort.
Acute liver failure (ALF) is a multifactorial process defined as loss of liver function that occurs rapidly usually with no pre-existing liver disease, unlike Acute on Chronic Liver Failure (ACLF) which requires underlying chronic liver disease (e.g. liver cirrhosis). SIRT may produce ALF/ACLF as well as subclinical liver injury: a significant and clinically irrelevant increase in total bilirubin after SIRT has been reported in different publications [41, 42]. On the other hand, Radioembolization-Induced Liver Disease (REILD) is a sinusoidal obstruction syndrome and defined by the appearance of jaundice and ascites 4–8 weeks after SIRT in the absence of tumor progression or bile duct occlusion. In general main liver complications do not result from the microembolic effect of SIRT but rather from radiation effects on the non-targeted liver tissue. Furthermore, radiation-induced blood vessel damage can result in clinically relevant liver toxicity after radioembolization in comparison with non-cirrhotic livers [43]. We recorded six cases (0.05%) of ALF/ACLF, which is lower than other reported data: Kennedy et. al reported 0.9%, and van Hazel et. al reported 1.3% in a combination arm with chemotherapy, both studies making no differences between ALF/ACLF and RAILD [36, 37]. We did not record any REILD cases. This seems logically given the common onset of RAILD (4–8 weeks after SIRT) and the fact that only a small portion of patients (1.1%) in our study stayed hospitalized for more than two weeks after SIRT and there were no cases of hospitalization beyond four weeks. There were no cases of radiation pneumonitis in our study, which is in line with data from the literature[37] and can furthermore be attributed to the observational period (hospital stay) in our study. It must be pointed out that due to the retrospective database evaluation we are unable to draw any causal link between the reported proportions and cannot exclude preexisting disorders or poor coding quality, but the reported frequencies are mostly in line compared to previously reported rates.
Population-based data about SIRT are largely lacking. During the observation period in our study, most SIRT interventions were performed in 2015. Individual cases increased steadily from 2012 to 2015 (mean increase per year: 10.5%) and decreased in the following years with the lowest count of SIRTs done in 2019. This finding can possibly be explained by increasingly strict assessment regarding cost coverage by the Medical Service of the Health Funds (MDK) in Germany and established alternative treatments (e.g. TACE) in most of the SIRT indications, but we are not able to prove this thesis. Based on our data and on the population numbers of the Federal States it is 19.7 times more likely to undergo SIRT in Saxony-Anhalt, compared to the federal state with the lowest rate, Brandenburg (Saxony-Anhalt: 61.1 SIRTs/year/1 million residents; Brandenburg: 3.1 SIRTs/year/1 million residents). These findings cannot be explained in terms of different regional tumor incidences (e.g. mean hepatic tumor (HCC + BTC)/100.000 residents 2012–2015 Saxony-Anhalt vs. Brandenburg: 8.1 vs. 7.7) [44]. The number of SIRTs performed in the New Federal States (26.4 SIRTs/year/1 million residents) was higher than in the Old Federal States (14.2), but the results were not statistically significant. Despite these mentioned large differences in the regional distribution, we could not identify structural or patient-related reasons for these findings: there was no correlation in terms of the overall population in the federal states, population density, number of university hospitals, average income, and the overall number of hospital beds. There was a correlation between the number of performed SIRTs and hospital beds per million residents (Pearson correlation coefficient: 0.43), as well as the mean average income (-0.4), but the results failed to show statistical significance. Other factors (e.g. local preferences and experience, different local availability and expertise), as well as historical reasons, may contribute to these findings.
We acknowledge some limitations of our study. First, it is a retrospective database evaluation unable to draw any causal link between performed SIRTs and the reported proportions. Furthermore, no information on coding quality in Germany is available and the database is not subject to systematic quality control between individual hospitals. Nevertheless, it can be considered that endpoints such as liver failure and death are little or not influenced by coding errors and may correctly represent medical practice. Another limitation is that the analysis of cases/SIRT procedures does not include individual circumstances, which may contribute to the observed results. Finally, our data do not reveal important SIRT parameters such as dose of radiation, pretreatment/subsequent therapies, and treatment selectivity.