The actual number of resin microspheres is approximately 30–60 times higher than glass microspheres per 3 GBq vial. Thus, radioembolization (RE) with resin microspheres exerts an embolization effect besides the radiation effect. This embolization effect can occasionally cause early back flow of the microspheres before application of the entire calculated dose. To avoid these adverse side effects, RE has to be terminated at an earlier time point. Measurement of the residual activity in the delivery box, which includes the v-vial, tube and catheter, to calculate the achieved target dose is often challenging. The aim of the current study was to establish a post-RE measurement method comparable to the glass microspheres method without unnecessary radiation exposure to the staff and risk of contamination. Methods: Two different measurements were performed. First, total radioactivity in the shipping vial was measured in an ion chamber and then it was put in the delivery box and the radiation was measured from a 30 cm distance from the centre of the box with a dosimeter. The required radioactivity was then transferred to the v-vial, and the shipping vial was measured again. After that, the v-vial was measured from the same distance from the centre of the box with dosimeter. Results:Altogether 62 times the shipping vial with different activities were measured with a significant positive correlation between the amount of the activity measured in the iron chamber and the radiation dose, measured with dosimeter (r² = 0.98; p< 0.001). There was also a strong positive correlation between these measurements of the v-vial (r² = 0.98; p< 0.001). Conclusion:With measurement of the residual activity in the delivery box using a dosimeter the percentage of the whole injected activity can be easily calculated. This facilitates the calculation of the actual, achieved target and non-target dose in those cases, where therapy had to be stopped because of eminent flow reversal or obstruction.

395 Objectives Heterogeneous tumor somatostatin receptor expression results in a wide intra-individual variation of tracer uptake in pre-therapeutic 68Ga-DOTATOC-PET/CT of pts with gastro-enteropancreatic neuroendocrine tumors (GEP NET) undergoing peptide receptor radionuclide therapy (PRRT). This study aims to assess the lesion-based relation of receptor mediated tumor uptake and functional response to PRRT. Methods 32 pts with metastatic GEP NET (12 pancreatic, 20 non-pancreatic) treated with 177Lu-octreotate (4 cycles, mean 7.9 GBq per cycle), 68Ga-DOTATOC-PET/CT at baseline and 3 mo after the last PRRT cycle, were included in the retrospective analysis. Clearly delineated tumor lesions were selected as target lesions (2-3/pt). SUVmax, SUVmean (automated segmentation, threshold of 50% of SUVmax), and respective tumor-to-liver ratios (TLR) of target tumor lesions to normal liver parenchyma were calculated. Response of each tumor lesion was evaluated according to the relative change in functional tumor volume (%ΔTVPET). The lesion-based association of baseline SUV parameter and functional response was tested using Spearman9s rank correlation analysis. Results 71 lesions were included into the analysis. Mean SUVmax and SUVmean at baseline were 28.1 ±15.9 and 13.6 ±5.1, respectively. ΔTVPET 3 months after completion of PRRT was 39.6±52.1 %. Baseline SUVmax and SUVmean showed a poor (p=0.614 and p=0.082, respectively), and TLR a better correlation (p=0.002 each) with lesion-based response (SUVmeanTLR, r=0.412; SUVmaxTLR, r=0.434). Among the other characteristics including the origin of primary, tumor volume at baseline and metastatic site, only pancreatic origin contributed to functional volume reduction (ΔVPET%: 70.1±29.5 in pancreatic vs. 27± 54.3 in non-pancreatic NET; p=0.041). Conclusions Lesion-based molecular response to PRRT is related to pretreatment uptake in somatostatin receptor PET quantified in the form of tumor-to-liver SUV ratios.

Summary[177Lu-DOTA0, Tyr3]-octreotate (177Lu-octreotate) in peptide receptor radionuclide therapy (PRRT) offers direct intra-therapeutic dosimetry. The aim of this study was to compare tumour and non-tumour parameters and assess intra-individual variations. Patients, methods: Retrospective analysis of 53 consecutive PRRT treatment cycles (mean activity of 7.53 ± 0.46 GBq 177Lu-octreotate, intended four cycles at intervals of 10–14 weeks, standard nephroprotection) in 27 GEP NET patients. Extended planar dosimetry with serial wholebody imaging on selected, non-superimposed tumour and non-tumour regions; liver (LM), bone (BM), and other (OM) metastases. The per-cycle variation was compared with posttreatment response (CT/MRI three months post-treatment, modified SWOG criteria). Results: Residence time in tumor lesions (133–147 h) exceeded that in kidneys (93 h). Tumour-to-kidney absorbed dose ratios ranged from 14 to 28 (LM, BM, OM). Intra-individual per-cycle dose variation was insignificant for kidneys, but significant for metastases (LM, BM, and OM; p < 0.05). The mean per-cycle decrease of tumour absorbed dose (_D/A0[%]) was linked to morphologic response after PRRT. A mean decrease of >20% was predictive of a partial or minor remission in all 11 evaluable patients, while absent significant dose reduction indicated stable or progressive disease in 4/5 patients. The dose decrease was unrelated to volume effects and also observed for BM. Conclusion: Besides confirmation of a favourable tumour-to-kidney parameter relation for 177Lu-octreotate, stepwise intra-lesional comparison seems to imply a prognostic impact of tumor dosimetry: The early per-cycle change _D/A0 between treatment cycles may predict the outcome after PRRT. Larger studies are needed to confirm this finding.

[177Lu-DOTA0,Tyr3]-octreotate (177Lu-octreotate) in peptide receptor radionuclide therapy (PRRT) offers direct intra-therapeutic dosimetry. The aim of this study was to compare tumour and non-tumour parameters and assess intra-individual variations. Patients, methods: Retrospective analysis of 53 consecutive PRRT treatment cycles (mean activity of 7.53 ± 0.46 GBq 177Lu-octreotate, intended four cycles at intervals of 10–14 weeks, standard nephroprotection) in 27 GEP NET patients. Extended planar dosimetry with serial whole-body imaging on selected, non-superimposed tumour and non-tumour regions; liver (LM), bone (BM), and other (OM) metastases. The per-cycle variation was compared with post-treatment response (CT/MRI three months post-treatment, modified SWOG criteria). Results: Residence time in tumor lesions (133–147 h) exceeded that in kidneys (93 h). Tumour-to-kidney absorbed dose ratios ranged from 14 to 28 (LM, BM, OM). Intra-individual per-cycle dose variation was insignificant for kidneys, but significant for metastases (LM, BM, and OM; p 20% was predictive of a partial or minor remission in all 11 evaluable patients, while absent significant dose reduction indicated stable or progressive disease in 4/5 patients. The dose decrease was unrelated to volume effects and also observed for BM. Conclusion: Besides confirmation of a favourable tumour-to-kidney parameter relation for 177Lu-octreotate, stepwise intra-lesional comparison seems to imply a prognostic impact of tumor dosimetry: The early per-cycle change ΔD/A0 between treatment cycles may predict the outcome after PRRT. Larger studies are needed to confirm this finding.

SummaryThe actual number of resin microspheres is approximately 30-60 times higher than glass microspheres per 3 GBq vial. Thus, radioembolization (RE) with resin microspheres exerts an embolization effect besides the radiation effect. This embolization effect can occasionally cause early back flow of the microspheres before application of the entire calculated dose. To avoid these adverse side effects, RE has to be terminated at an earlier time point. Measurement of the residual activity in the delivery box, which includes the v-vial, tube and catheter, to calculate the achieved target dose is often challenging. The aim of the current study was to establish a post-RE measurement method comparable to the glass microspheres method without unnecessary radiation exposure to the staff and risk of contamination. Methods: Two different measurements were performed. First, total radioactivity in the shipping vial was measured in an ion chamber and then it was put in the delivery box and the radiation was measured from a 30 cm distance from the centre of the box with a dosimeter. The required radioactivity was then transferred to the v-vial, and the shipping vial was measured again. After that, the v-vial was measured from the same distance from the centre of the box with dosimeter. Results: Altogether 62 times the shipping vial with different activities were measured with a significant positive correlation between the amount of the activity measured in the iron chamber and the radiation dose, measured with dosimeter (r2 = 0.98; p < 0.001). There was also a strong positive correlation between these measurements of the v-vial (r2 = 0.98; p < 0.001). Conclusion: With measurement of the residual activity in the delivery box using a dosimeter the percentage of the whole injected activity can be easily calculated. This facilitates the calculation of the actual, achieved target and non-target dose in those cases, where therapy had to be stopped because of eminent flow reversal or obstruction.

1186 Objectives Renal radiation during peptide receptor radionuclide therapy (PRRT) may result in glomerular damage and reduction of glomerular filtration rate (GFR) and eventually may lead to renal failure years after PRRT. Serum creatinine and creatinine clearance allow only an approximate estimation of the GFR. The aim of this study was an accurate assessment of long -term changes in GFR using the 99mTc-DTPA clearance. Of particular interest was the impact of pre-existing risk factors. Methods GFR of 75 patients with gastroenteropancreatic neuroendocrine tumours (GEP-NET) treated with 177Lu-octreotate (4 intended cycles at 3 monthly intervals and mean activity of 7.9 GBq per cycle) were retrospectively analyzed. A minimum follow-up of 12 months was required for patient inclusion. Mean follow-up duration was 20 months (range 12-48 months), with a median of 5 GFR measurements per patient. The change of GFR was analyzed by linear curve fit. Potential risk factors including DM, HTN, previous chemotherapy and decreased renal function at baseline were analyzed regarding impact on the renal function loss. Comparability of nephrotoxicity determined by 99mTc-DTPA clearance and serum creatinine (using Common Terminology Criteria for Adverse Events v3.0) was also investigated. Results A reduction of >2 ml / year in GFR was observed in 31 (42%) and > 10ml/year in 16 patients (22%). The mean and median decline in GFR was 12 and 7 ml per year respectively. However, only one patient developed serious nephrotoxicity (CTCAE grade III). Eleven patients showed an increase of > 10 ml/year in GFR following PRRT. Nephrotoxicity graded according to serum creatinine (CTCAE) was not compatible with GFR in 17% of the assessments and was underestimated by serum creatinine in 12% of patients. None of the investigated risk factors contributed to a reduction in GFR. Conclusions Severe nephrotoxicity after PRRT with 177Lu-octrotate is rare (1.3 % of patients). However, mild renal function impairment (> 2 ml/year) is common and best monitored with 99mTc-DTPA clearance

PURPOSE:Selection of candidates for peptide receptor radionuclide therapy (PRRT) is increasingly based on receptor positron emission tomography (PET) imaging, including the common tracer 68Ga DOTATOC. However, no studies have yet compared standardized uptake values (SUVs) and absorbed doses in this field. MATERIALS AND METHODS:We retrospectively analyzed a consecutive cohort of 21 patients with 61 evaluable tumor lesions undergoing both pretherapeutic 68Ga DOTATOC-PET/CT (Biograph Duo [Siemens Medical Solutions, Erlangen, Germany]; PET acquisition, 75.3 ± 15.4 minutes postinjection; 117.3 ± 33.9 MBq 68Ga DOTATOC) and PRRT with Lu octreotate (7.47 ± 1.39 GBq; intratherapeutic tumor dosimetry with serial whole-body scans; 1, 2, and 4 days postinjection) at our institution. SUVs were compared with the tumor-absorbed doses per injected activity (D/A0) of the subsequent first treatment cycle. RESULTS:The correlation of SUV and D/A0 was r = 0.72 (SUVmean) and r = 0.71 (SUVmax), both P < 0.001. Pancreatic origin and hepatic localization were associated with higher D/A0, and chromogranin A level and Ki-67 index had no influence on SUV or D/A0. High-SUV lesions (SUVmean >15; SUVmax >25) resulted in high D/A0 (>10 Gy/GBq) in 66.7% to 70.8% and low D/A0 (<5 Gy/GBq) in only 8.3% to 12.5% on subsequent PRRT. The mentioned low D/A0 range, on the other hand, was achieved by all lesions with SUVmean <7 or SUVmax <9. CONCLUSIONS:Somatostatin receptor PET imaging may predict tumor-absorbed doses. The ability to indicate insufficient target irradiation by a low SUV could aid in selection of appropriate candidates for PRRT. However, larger series are needed to confirm and validate these initial findings.

92 Objectives The absence of γ emission limits accurate scintigraphic dosimetry after radioembolization with Y90 microspheres (RE).SPECT/CT is currently recommended to evaluate the radioactive distribution and to rule out any extrahepatic shunting using bremsstrahlung(BS).Y90-PET/CT is recently being studied as an alternative modality.The aim of this study is to find a feasible reconstruction method for Y90-PET/CT and to compare the accuracy of PET/CT and SPECT/CT for the evaluation of RE especially considering the extrahepatic region. Methods 18 Y90-PET/CT (Biograph 2 Siemens) and BS SPECT/CT (Symbia T2, Siemens)scans were performed for 15 patients(6 men, mean age:61y)24 hours(median time)after RE(mean dose:1.8 GBq).BS SPECT/CTs were acquired using a MEGP collimator with a very broad energy window of 55-250. Additionally, each patient underwent a 90Y-PET/CT scan consisting of 1-2 bed positions with an acquisition time of 20 min/bed position.Tomographic images were reconstructed by both iterative and filtered back projection.Different reconstruction parameters were tested to find the optimal reconstruction method.The obtained images of both modalities were compared with each other and with FDG-PET/CT or MRI images. Results Considering intrahepatic tracer distribution Y90-PET/CT scan showed a good correlation with corrected BS SPECT/CT after iterative reconstruction with 128x128 matrix and 8 mm FMWH,while more than 4 iterations didn’t improve the quality of images.The main limitations of PET/CT in comparison to SPECT/CT were the artifacts due to low counts hindering the evaluation of the extrahepatic region in all patients(100%)and its convenience for the patients due to the much longer scan time. Conclusions Post therapy distribution of Y90microspheres in liver can be easily evaluated with PET/CT using iterative reconstruction. However,current quality of PET/CT does not allow us to rule out extrahepatic non target application of microspheres and therefore BS SPECT/CT is still the modality of choice for this purpose

Activated mitogen-activated protein kinase MAPK cascade leading to ERK1/2 phosphorylation is expressed in the majority of glial neoplasms and negatively correlates with survival time of patients. Here we show that ERK1/2 kinases are constitutively activated in glioma cell lines and stem cell-enriched primary cultures of glioblastoma. Pharmacological targeting of the activated MEK/ERK1/2 module with the MEK inhibitor U0126 attenuates cell cycle progression (11 out of 11 cell lines), impairs single (7 out of 10) and collective cell migration (9 out of 11) and abolishes single cell emigration from monolayers (4 out of 9). Attacking the activated MEK/ERK1/2 module thus partially blocks the tumorigenic potential of glial cancer cells on different levels and strongly suggests the application of combination molecularly targeted therapies to interfere more efficiently with glial tumor development and progression.

An angiographic evaluation combined with 99mTc-macroaggregated albumin (Tc-MAA) scanning should precede the treatment of any selected candidates for radioembolization (RE) of the liver. If the tumours in one liver lobe have not been targeted in the test angiogram, it should be repeated. However, in a few cases treatment of one liver lobe or at least some segments is safe and feasible and performing a repeated test angiogram with Tc-MAA (Re-MAA) in a separate session leads to more radiation exposure and could be time consuming. Our aim was to evaluate the feasibility of concurrent RE of a part of the liver and therapy planning for another region by simultaneous injection of the Tc-MAA and 90Y-microspheres in two different locations in the therapy session. Tc-MAA and bremsstrahlung (BS) single photon emission computed tomography (SPECT)/CT were performed separately in an effort to distinguish between the distributions of these two different radiopharmaceuticals.

Purpose: In patients with drug-refractory focal epilepsy, nonlesional magnetic resonance imaging (MRI) or discordant data of presurgical standard investigations leads to failure generating a sufficient hypothesis for electrode implantation or epilepsy surgery. The seizure-onset zone can be further investigated by subtraction ictal single-photon emission computed tomography (SPECT) coregistered to MRI (SISCOM). This is an observational study of a large consecutive cohort of patients undergoing prospective SISCOM to generate hypothesis for electrode implantation or site of epilepsy surgery.Methods: One hundred seventy-five consecutive patients undergoing presurgical evaluation with either nonlesional MRI or discordant data of standard investigations preventing the generation of hypothesis for seizure onset were evaluated with SISCOM.Results were compared to gold standard for seizure onset detection, either electrocorticography (ECoG) and/or postoperative outcome.Key Findings: One hundred thirty patients had successful SPECT injection. Hypothesis for electrode implantation/site of surgery was generated in 74 patients. Forty patients had gold standard comparison. Twenty-eight patients underwent resective surgery. SISCOM was concordant to site of surgery in 82%. An additional 12 patients underwent invasive EEG monitoring but were not suitable for surgery. SISCOM was concordant multifocal in 75%. Two years postsurgical follow-up of 26 patients showed favorable outcome in 22 (Engel class I and class II). Significance: SISCOM is a highly valuable diagnostic tool to localize the seizure-onset zone in nonlesional and extratemporal epilepsies. Outcome in this patient group was unexpectedly good, even in patients with nonlesional MRI. The high correlation with ECoG and site of successful surgery is a strong indicator that outcome prediction in this patient group should be adapted accordingly, which may encourage more patients to undergo electrode implantation and subsequent successful surgery.Statistical analysis showed that SISCOM with shorter duration of seizures, focal seizures, and lesional MRI was more likely to generate implantation hypothesis.

Background: As single-head data acquisition for thallium-201 myocardial SPECT is a frequently used method mainly in the outpatient medical care as well as in smaller hospitals, comparison to dual-head data collection is a still discussed issue mainly with regard to quality control and -assurance.Methods: A total of 1334 patients undergoing thallium-201 myocardial SPECT for diagnosis of myocardial ischemia and/or viability have been retrospectively analyzed. In 554 patients, a single-head gamma camera (360 degrees rotation) has been applied, whereas a dual-head gamma camera (180 degrees rotation) has been used in 780 patients. Four hundred twenty-six patients received both myocardial SPECT as well as coronary angiography. The diagnostic value of both applied acquisition techniques has been analyzed.Results: Regarding myocardial viability, positive predictive value for the diagnosis of myocardial scar tissue was significantly higher for dual-head-as compared to single-head acquisition. Among the 426 patients undergoing diagnosis of myocardial ischemia, significant differences have only been found with regard to specificity being higher in the single-head acquisition. Diagnosis of myocardial ischemia related to a distinct myocardial perfusion region showed a significantly higher sensitivity of dual-head acquisition for the left anterior descending perfusion area, whereas specificity was significantly higher for single-head acquisition.Conclusions: Our results indicate a beneficial effect of dual-head data collection with regard to sensitivity of the diagnosis of myocardial ischemia. In contrast, single-head data acquisition was superior with regard to specificity. However, it is justified to preferably apply dual-head data collection in clinical routine due to the shorter acquisition time leading to an evident time benefit of this acquisition technique. (C) 2007 Elsevier Inc. All rights reserved.

1667 Objectives: Assessment of response and tolerability of receptor-mediated radionuclide therapy in gastroenteropancreatic neuroendocrine tumors (GEP-NET) using Lu-177-DOTA octreotate with prolonged intervals. Methods: We analyzed n=25 consecutive pts with inoperable GEP-NET (mean age 62 y [37-77]) and ≥ 2 treatment cycles (each with 7.4 GBq Lu-177-DOTA octreotate); minimum interval of 3 mo, renal protection with co-infusion of lysine-arginine-solution; dosimetry with serial post-therapeutic whole body imaging, Picker Prism 2000 with MEGP collimator, energy peaks at 113/208 keV. At baseline and during follow-up regular lab examinations including complete blood counts, hepatic and renal function tests, tumor markers including chromogranin A and NSE, monitoring of GFR (Tc-99m-DTPA clearance). Radiological response was assessed by CT and/or MRI using RECIST. Toxicity was classified according to CTC. Results: Preliminary response was as follows: all GEP-NET (n=25; mean 2.5 [2-4] courses per pt.) PR 9/25 (36%), MR 4/25 (16%), NC 10/25 (40%), PD 2/25 (8%) pts.; pancreatic NET (P-NET; n= 12, mean age 61 y [37-77], mean 2.6 courses per pt.) PR 58%, MR 25%, NC 16%; non-pancreatic GEP-NET (GE-NET; n=13, mean age 63 y, 42-75; mean 2.5 courses per pt.) PR 15%, MR 8%, NC 62%, PD 15%. Treatments were overall well tolerated, nausea was mostly transient and self-limiting (CTC 1-2), in one case (small bowel NET) a post-therapeutic GI-bleeding originating from the site of the primary was noted, no other relevant GI-toxicities were seen, transient hematotoxicity grade 3

1673 Objectives: The repeated radionuclide therapy of bone metastases of the prostate with Rhenium-188 HEDP has been proved to be advantageous. The aim of this study was to determine the irradiation dose in bone metastases and other compartments of the body (soft tissue, normal bone, bone marrow) that underwent multiple treatments with Re-188 HEDP. Methods: 81 patients underwent posttherapeutic whole body scans (immediately p.i., after 1.5 h, 18 h, 24 h, 42 h). Using the ROI-technique several metastases, normal bone, soft tissue and whole body were evaluated. The doses were calculated after single and repeated radionuclide therapy. So far 55 therapies at 19 patients with prostate cancer and multifocal bone metastases (2-6 therapies/patient; 14 patients with more than 2 therapies; mean applied activity/therapy: 3.2 GBq Re-188 HEDP) were analyzed. Results: The mean dose in the metastases per patient was 11.3 Gy after single injection. The cumulative dose after repeated therapy (mean 2.9) was 34.7 Gy. The intraindividual comparison of the doses of the metastases from 1st and 2nd injection gave us the following results: 6 / 13 patients showed a higher / equal dose after the 2nd therapy (no patient with lower dose). The result of the same comparison from 2nd and 3rd injection was: 2 of 7 patients showed higher or equal dose after the 3rd therapy; in 5 patients the dose was lower. The mean soft tissue dose per patient was 0.5 Gy after each injection. Conclusions: Following these results a repetitive therapy of at least 3 injections is meaningful to obtain a higher cumulative dose in the bone metastases. Further evaluations have to show, if up to 6 injections are reasonable.

SummaryAim: Dosimetry in 131I-lipiodol therapy for hepatocellular carcinoma (HCC) in the hitherto largest existing patient cohort. Patients, methods: 38 courses of intra-arterial 131I-lipiodol therapy with a total activity up to 6.7 GBq were performed in 18 patients with HCC. Liver and tumour volume were measured by computed tomography (CT) and 131I activity by scintigraphy on day 3, 6, 14, 28 and 42 after injection. Lipiodol deposition in tumour nodules as shown by CT rendered definite attachment to scintigraphic data possible. The radiation dose in tumour nodules, liver and lungs was calculated according to the MIRD concept and the tumour dose related to pre-therapeutic tumour volume, response and survival. Results: Mean tumour dose was 23.6 ± 3.6 Gy (14.2 ± 2.1 mGy/MBq) with maximal 162 Gy (90.1 mGy/MBq) after one and 274 Gy after three courses. The dose to nontumourous liver was 1.9 ± 0.2 Gy (1.2 ± 0.1 mGy/MBq) and the mean dose ratio of tumour / nontumourous liver 11.1 ± 1.7 (max. 82). The pulmonary dose was 25.9 ± 1.8 mGy (16.3 ± 1.2 μGy/MBq) and therefore much lower. There was a reciprocal relation between tumour dose and pretherapeutic tumour volume. Tumour dose had no effect on response or survival. Conclusion: High radiation doses are particularly in small tumour nodes achievable but not necessarily related to tumour response. The dose of non-tumourous liver and lungs is much lower.

SummaryAim: To evaluate the efficacy and tolerance of iodine- 131-lipiodol (131I-lipiodol) for hepatocellular carcinoma (HCC) in German long term patients and comparison with medically treated controls. Patients, Methods: 38 courses of intra-arterial 131I-lipiodol therapy with a total activity up to 6.7 GBq were performed in 18 patients with HCC (6 with portal vein thrombosis). Liver and tumour volume and lipiodol deposition were measured by computed tomography and 131I activity by scintigraphy. Therapeutic efficacy was determined by tumour volume change and matched-pairs analysis in comparison to medically (i.e. tamoxifen or medical support) treated patients. Results: Tumour volume decreased in 20/32 index nodules (63%) after the first course. Repeated therapy frequently resulted in further tumour reduction. Overall response to treatment was partial in 11 nodules, minor response in 4 nodules, and disease was stable in 12 and progressive in 5. Significant response was associated with pretherapeutic nodule volume up to 150 ml (diameter of 6.6 cm). Survival rate after 3, 6, 9, 12, 24 and 36 months was 78, 61, 50, 39, 17, and 6%. Matched-pairs analysis of survival revealed 131I-lipiodol to be superior to medical treatment. The most important side effect was a pancreatitis-like syndrome whereas overall tolerance was good. Conclusion: The long term results confirm that HCC therapy with 131I-lipiodol is effective and probably superior to medical treatment. Tumour nodules of up to 6 cm diameter are well suited for this therapy even in the presence of portal vein thrombosis.