HCC and tumor dose
With the exception of one study’s negative results, those of the five other available studies were quite concordant, clearly demonstrating there to be a dose/response relationship (Tables
1).
Table 1
Studies with MAA based tumor dose evaluation in HCC
Product | resin | resin | resin | glass | glass | glass |
Dosimetry | NAa
| MIRD | MIRD | MIRD with BEDs | MIRD | MIRD |
Nb patients | 19 | 71 | 10 | 52 | 36 | 71 |
Nb lesions | NA | NA | NA | 65 | 58 | NA |
Lesion size (cm) | NA | NA | NA | 5.6 | 7.1 | 7.1 |
Prior therapy (%) | NA | NA | Yes (50) S (NA) CE (NA) Other (NA) No (50) | Yes (28.9) S (15.5) CE (0) Other (13.4) No (71.1) | Yes (42) S (13.8) CE (25) Other (2.7) No (58) | Yes (51) S (22.5) CE (18.3) Other (32.6) No (49) |
Number of radioembolization (%) | NA | 1 (77.9) 2 to 5 (22.1) | 1 (100) | 1 (89.6) 2 (10.4) | 1 (61) 2 (39) | 1 (69) 2 (31) |
Response Evaluation | RECIST1.1 | WHO | RECIST1.1 | EASL | EASL | EASL |
Time of evaluation | 6 w | NA | NA | 3 m | 3 m | 3 m |
Dose/response relationship | NA | YES | Probablyb
| YES | YES | YES |
threshold dose (Gy) | NA | 225 | NA | 257 | 205 | 205 |
Impact on survival | NA | NA | NA | NA | YES | YES |
The study of Kucuk et al. [
40], using resin microspheres in a group of 19 patients, was the only negative outcome study in the HCC setting. The response rate that they recorded at 3 months, defined according to RESIST 1.1 criteria, did not significantly differ between hypoactive or hyperactive lesions, recorded at 40 % and 58 %, respectively (
p = 0.51). However, this study included no dosimetric evaluation; only a qualitative evaluation of the MAA uptake.
Using the partition model, Ho et al. [
33] were the first to report a link between tumor dose and response. In this study, involving 71 HCC patients treated with resin microspheres, the response rate was 37.5 % for lesions with a tumoral dose >225 Gy versus only 10.3 % if the tumoral dose was ≤225 Gy (
p <0.006). Nevertheless, overall survival did not statistically differ depending on the cumulative tumoral dose (< or ≥300 Gy,
p = 0.35). In a second study evaluating the resin microsphere method, Kao et al. [
41] also reported interesting findings, though this was a preliminary study and involved only eight evaluable patients. Still, all the responding lesions had received a tumoral dose >91 Gy. Owing to all the lesions having responded to treatment, a threshold tumor dose could not be clearly identified in this study.
Chiesa et al. [
12] and Mazzaferro et al. [
6] conducted studies using glass microspheres, with the former offering preliminary results on 48 patients [
12] and the latter offering full results [
6] on 52.
A clear dose/response relationship was identified using a dosimetric evaluation based on BEDs at the voxel level. Non-responding lesions had received a median tumor dose of only 199 Gy, compared to 431 Gy for the responding ones (
p <0.0001) [
12]. For response prediction, defining a non-responding lesion with a TD <257 Gy as the true negative and responding lesion with a TD >257 Gy as the true positive, the threshold tumor dose of 257 Gy exhibited a sensitivity of 85 % and specificity of 70 % [
12].
In a preliminary study [
27], our group also observed a strong dose/response relationship in 36 patients with 58 evaluable lesions of relatively large size (mean size: 7.1 cm). The mean TD was 372.7 ± 142.0 Gy for the 45 responding lesions and only 153.8 ± 80.8 Gy for the 13 non-responding ones (
p <0.0001). None of the lesions receiving a TD below 205 Gy responded, whereas only five receiving >205 Gy did not respond. This threshold TD of 205 Gy was thus confirmed to be predictive of response, with a sensitivity of 100 % and accuracy of 91 %. According to multivariate analysis, TD was the only parameter that correlated with response (
p = 0.019). In this study, TD also affected survival. Progression-free survival (PFS) was only 5.2 months when the TD was <205 Gy versus 14 months (
p = 0.0003) with higher doses. Overall survival (OS) was 9 months when the TD was <205 Gy
versus 18 months (
p = 0.0322) with a TD of 205 Gy or higher.
These findings, pertaining to a 205 Gy threshold TD and correlation between TD and survival, were confirmed in a recent study involving 71 patients [
7], in which the concept of a personalized dosimetric approach, including treatment intensification when necessary, was also described. The patients who received treatment intensification were administered an ILD ≥150 Gy, contrasting with the 120 ± 20 Gy delivered in the classical approach. In this concept, the HILD was <120 Gy and the LD was typically <30 Gy for one treatment or <50 Gy for several. In total, 38 % of patients received treatment intensification. The response rates were significantly higher with the personalized dosimetric approach than with the standard dosimetric approach, recorded at 86 % versus only 55 %, respectively (
p = 0.01).
This intensification concept appears to be of particular value for PVT patients. Personalized dosimetry, with treatment intensification where necessary, has been described in a study involving 41 PVT cases [
42]. In this trial, 37 % of patients received treatment intensification. A high response rate of 85 % was achieved without causing any increase in liver Grade ≥ III permanent toxicity (6 % versus 12 % in the non-boosted patients, ns). The TD was found to exhibit a highly significant impact on OS, which was 4.3 months (3.7-5 months) versus 18.2 months (8.5–28.7 months) for patients with a TD below 205 Gy or over 205 Gy, respectively (
p = 0.005). Those with a TD ≥205 Gy and good PVT targeting (n = 36) exhibited an OS of 20.9 months. The objective median OS was not reached, though it was longer than 24.5 months and significantly longer (
p = 0.0493) for the five patients who underwent lobar hepatectomy.
Several studies have produced disappointing results in this context, yet there are some methodological concerns that could account for this (Table
2). For example, only MAA uptake qualitative evaluation was reported on some occasions, with no mention of type of dosimetric approach [
43,
44]. Also, the morphological response evaluation used was at times inappropriate, not including necrosis or hypervascularization evaluation [
43,
44], and the means of evaluating delay of response was also insufficient, consisting of 4–6 weeks instead of 3 months [
19]. Five studies produced disappointing results, four evaluating resin microspheres and one glass microspheres.
Table 2
Studies with MAA based tumor dose evaluation in metastatic disease
Product | resin | resin | resin | resin | glass | resin | resin |
Dosimetry | NA a
| NA b
| NA c
| MIRD | MIRD | Monte carlo | MIRD |
Nb patients | 20 | 31 | 58 | 66 | 13 | 8 | 25 |
Nb lesions | NA | 225 b
| NA | 435 | 91 | 39 | NA |
Lesion size (cm) | NA | NA | NA | 3.4 | 3.5 | 5.7 | NA |
Prior therapy (%) | NA | NA | NA | YES (chemo) (100) | YES (chemo) (100) | YES (chemo) (100) | YES (chemo) (100) |
Concomitant Chemotherapy | NA | NA | NA | NA | NA | YES | NA |
Number of radioembolization (%) | NA | NA | 1 (100) | NA | 1 (84.6) 2 (15.4) | NA | 1 (100) |
Response Evaluation | NA | NA | WHO | RECIST1.1 | FDG PET | FDG PET | RECIST1.1 |
Time of evaluation | NA | NA | 2 m | 3 m | 4-5 weeks | 6 weeks | 3 m |
Dose/response relationship | NA | NA | NA | NA | NA | YES | YES |
threshold dose (Gy) | NA | NA | NA | NA | NA | NA | 44.2 |
Impact on survival | NA | NA | NA | NA | NA | NA | YES |
Knesaureck et al. [
45] reported a comparative study of MAA and resin microsphere uptake in a group of 20 patients (lesion size not available). A strong correlation was demonstrated for some patients, while others showed poor correlation, with Spearman’s rank values of between 0.451 and 0.818.
Wondergem et al. [
23], also comparing MAA and resin uptake in 225 hepatic segments, reported disappointing results in 31 patients, primarily involving metastasis. Differences of >10 %, >20 %, and >30 % in the mean activity per milliliter was found in 68 %, 43 %, and 32 % of the 225 segments analyzed, respectively. Tumor burden significantly influenced these differences, with smaller discrepancies observed for segment involvement >25 %.
The 2005 Dhabuwala et al. [
44] study reported no correlation between qualitative MAA uptake and response in 58 patients treated by resin microspheres. Furthermore, CT response rates at 3 months did not, in fact, differ between patients with high MAA uptake (n = 37) and those with equivocal or low MAA uptake (n = 21). Nevertheless, tumor response was only evaluated based on lesion size changes. In other studies, the MAA and microsphere injections were not performed in comparable situations, with angiotensin II injected only prior to microsphere administration. Due to its recognized capacity to increase tumoral vascularization [
46,
47], MAA was unable to accurately predict microsphere biodistribution in this study.
In their large study involving 66 patients and 435 lesions, Ulrich et al. [
43] evaluated the use of resin microspheres and also found no correlation between MAA qualitative uptake and response. The mean tumoral size was relatively small (3.4 cm), with response evaluated solely using lesion size changes. At 3 months, 290 lesions were evaluable according to RECIST 1.1 criteria. In total, 22 % of the responding lesions exhibited low MAA uptake and 21.7 % of the non-responding lesions high MAA uptake.
Van de Wiele et al. [
19] conducted another study, this time using glass microspheres, and also found there to be no correlation between the accuracy of MAA quantification and response, as the responding lesions exhibited a mean microsphere activity of 1.95 MBq/cc compared to 1.90 MBq/cc for non-responding ones (
p = 0.92). However, response was evaluated very early in their study, namely at 4–5 weeks, with the mean lesion size being quite small (3.5 cm) and MAA uptake very low, with a mean uptake of only 1.5 % per lesion. This last finding clearly demonstrates the significantly different vascular behavior of metastasis in comparison with HCC, given that the mean reported MAA uptake in HCC was 32.8 % [
27].
Two studies published positive data on this issue.
Flamen et al. [
28] conducted a preliminary study evaluating resin microspheres involving eight patients and 39 lesions with a mean size of 5.7 cm. They reported a strong correlation between the tumoral absorbed dose, evaluated on
99mTc-MAA SPECT imaging, and the FDG PET response to
90Y-resin microspheres in colorectal metastatic disease using an appropriate dosimetric approach, namely the Monte Carlo simulation. The median simulated absorbed dose was 20 Gy for the 20 poorest responding lesions and 46 Gy for the 19 responding ones (
p <0.001).
More recently, Lam et al. [
48] reported very interesting results involving 25 metastatic patients treated by resin microspheres. In order to avoid bias, only those with identical MAA and microsphere injection sites, as well as no injection failures, were analyzed, with response evaluated at 3 months according to the RECIST 1.1 criteria. Their chosen approach was original, based on both MAA scintigraphy and
99mTc-sulfur colloid (SC) liver scintigraphy for the segmentation between tumor and surrounding liver tissue. Furthermore, the authors used and tested different fixed thresholds for both MAA and SC volume delineation. This consisted of tumor and healthy liver segmentation, with the aim of suppressing manual segmentation, which would be of particular interest in cases of multifocal disease, just as that encountered in metastatic ones. Their dosimetric approach was based on the MIRD concept. The mean TD was 82.7 ± 23.9 Gy for responders
versus 31.0 ± 10.9 Gy for non-responder
p (p <0.001). A threshold tumoral dose of 44.2 Gy correlated with response (
p <0.001). The authors made a very interesting observation that the TD exhibited a clinical impact on survival. The median OS of patients who received a TD of over 55 Gy was 32.8 months, whereas those who received less than 55 Gy had a median survival of only 7.2 months (
p <0.05). The TD was the only parameter that correlated with survival (
p <0.01).