Background

Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer, with the third highest solid tumor mortality rate after lung and colon cancer [1]. In the USA, it is estimated that in 2020, 42,810 new cases of HCC were diagnosed, with a 5-year survival rate of 18% [2]. In early-intermediate HCC, liver resection or liver transplantation may be appropriate treatment options [3], but approximately 70–80% of patients with HCC are not diagnosed early enough to benefit from these potentially curative treatments [4, 5].

Several treatment strategies are available for patients with unresectable HCC. One such strategy is selective internal radiation therapy (SIRT), also referred to as radioembolization. SIRT is a technique that selectively deposits yttrium-90 (90Y) microspheres into the hepatic vasculature to deliver a lethal dose of radiation to the tumor(s). Surrounding healthy liver tissue can be spared by selectively delivering the microspheres into specific Couinaud segment(s) or the tumor vasculature itself. SIRT exploits the tumor’s hypervascularity relative to healthy tissue, delivering safe levels of radiation to normal tissue while achieving a selective tumoricidal dose. Relative to bland embolization or chemoembolization, SIRT minimizes potential macroembolic effect. Instead, SIRT leads to partial occlusion (microembolization) of the vascular supply to the tumor while the beta-radiation emitted from 90Y induces DNA damage in tumor cells [6, 7].

SIRT has the potential to not only delay local progression but also to downstage the disease to allow resection or transplantation if the disease meets Milan, or similar, criteria [8,9,10]. Previous randomized clinical trials of first-line 90Y resin microspheres (SIR-Spheres®, Sirtex Medical) for advanced HCC did not meet their primary endpoints for a number of reasons that include: (1) inadequate dosimetry-treatment planning; (2) poorly controlled patient populations; and (3) no post-treatment confirmation of absorbed dose [11,12,13]. These trials were initiated and conducted before the development of our current understanding of radiation-dose–response following SIRT, and before the widespread acceptance of personalized dosimetry for SIRT [14].

To overcome the limitations of these trials, we initiated the Duration Of Objective Response with arterial Yttrium-90 (DOORwaY90) study of 90Y resin microspheres, which utilizes a personalized dosimetry approach using the partition model and is recruiting patients with unresectable HCC (Barcelona Clinic Liver Cancer [BCLC] stage A, B1, B2 and C). All eligible patients receive 90Y resin microspheres using either a selective or lobar approach.

Methods/design

Study design

The DOORwaY90 trial is a pivotal, prospective, multicenter, open-label, single-arm study. The objective is to evaluate the safety and effectiveness of 90Y resin microspheres as first-line treatment for unresectable HCC (BCLC stages A, B1, B2, and C).

All relevant institutional and local ethics committee approvals were gained before commencing the study (Additional file 1). The study is being conducted in 14 investigational sites in the USA and aims to enroll approximately 100 patients (Fig. 1). No single site will enroll more than 20% of the study population. Written informed consent is obtained from all participants (or their legally authorized representatives) prior to participation in this study.

Fig. 1
figure 1

Study flow diagram

Investigators and study sites are required to permit study-related monitoring, audits, Institutional Review Board (IRB) review, and regulatory inspection(s) and provide direct access to source data/documents. This study will have oversight by a Data and Safety Monitoring Board (DSMB), which will review summarized safety data (including enrollment, protocol deviations, and adverse events) and will use stopping rules as defined in the DSMB Charter. The sponsor may also terminate or suspend the study if the DSMB stopping rules have been met. Sites must submit any protocol amendments to the IRB and are required to forward a copy of the written approval to the CRO. The final dataset will be available to the sponsor/CRO and IRB.

Important aspects of the trial include: (1) personalized dosimetry for treatment planning; (2) centralized review of the treatment plans and verification scans; (3) use of an independent core laboratory for response assessment; (4) a specific patient population that would benefit from locoregional treatment; (5) potentially curative treatment; (6) confirmation of absorbed dose delivered to the tumor; and (7) assessment of long-term toxicity.

Patients

Adult (≥ 18 years of age) patients enrolled into this study have a diagnosis of HCC BCLC stage A, B1, B2, or C with a maximal single lesion size of ≤ 8 cm and a sum of maximal tumor diameters of ≤ 12 cm, with the entire tumor burden expected to be treatable within the perfused volume. Patients cannot be considered suitable for treatment by resection or ablation at the time of study entry. For safety considerations, ≥ 33% of the total liver volume (body surface area-based) must be disease free and not treated with 90Y [15]. All inclusion and exclusion criteria are summarized in Table 1.

Table 1 Patient eligibility criteria for DOORwaY90 trial

Treatment

Pre-SIRT

The methods of catheter angiography and hepatic arterial mapping for SIRT are well established and reviewed elsewhere [16]. Briefly, angiography is performed to identify the appropriate catheter position for administration of the microspheres and to visualize any hepatic or gastrointestinal vessels that may result in non-target dose deposition. If found, these vessels must be successfully circumvented before proceeding with SIRT, either by adjusting the catheter placement or by performing prophylactic embolization. After this evaluation, technetium-99m (99mTc)-macroaggregated albumin (MAA) is administered via the catheter at the planned treatment position as a surrogate for the 90Y microspheres. The biodistribution of the 99mTc-MAA is assessed via planar scintigraphy of the chest and abdomen and single photon emission computed tomography/computed tomography (SPECT/CT) of the liver.

Target and non-target tumor compartments and the treated non-tumoral liver tissue compartments are segmented, and volumes and 99mTc-MAA uptake determined. The patient-specific target tumor-to-normal ratio (TNR) that is needed as input for the partition model is then calculated. The catheter tip during 99mTc-MAA injection would typically be placed at the same anatomical position from which 90Y resin microspheres will be administered. The exception is when segmental treatments are planned, in which case, lobar 99mTc-MAA injection may be performed, at the discretion of the clinician. When a segmental treatment (number of segments ≤ 2) is planned, the TNR value is then set to 1.0 (i.e., plan as a single compartment), and the minimum planned dose to the segment is > 150 Gy. Multi-vessel treatments are permitted. However, a cone-beam computed tomography (CBCT) or in-room fan beam CT should be performed to adequately determine the perfused volume. 99mTc-MAA is administered at each location; exceptions include segmentectomy cases. Clinical justification for adjustment or alteration of catheter position between sessions is documented. All pre-SIRT images and contours are assessed jointly by the treatment physician and the Eligibility Review Committee to confirm treatment planning dosimetry and approach based on the dosimetry principles and constraints outlined next.

Treatment plan dosimetry

CBCT or in-room fan beam CT is used for volume calculation. 3D volumetric software is used to segment the total liver volume (TLV), the volume of perfused normal liver parenchyma (NLp) and the treated tumor (TT) compartments in the liver, and to determine their volume (in ml) and the total 99mTc-MAA counts in the NLp and TT compartments. Soft-tissue density of 1.06 g/ml is used to determine the mass (in kg) of the TT compartment (Mtumor) and the mass of the perfused normal liver compartment (MNLp). The patient-specific tumor uptake ratio, R, is calculated (using MIM Maestro) as:

  • R = tumor compartment (total 99mTc-MAA counts/total volume of interest)/perfused normal liver compartment (total 99mTc-MAA counts/total volume of interest)

Based on partition model equations [17], the administered activity (Aadmin) required to meet the intended mean dose to the tumor compartment, Dtumor, is calculated according to the formula:

  • Aadmin [GBq] = Dtumor [Gy] × Mtumor [kg]/(49.7 [Gy-kg/GBq] × (1-LSF) × R × Mtumor/(MNLp + R × Mtumor)), where LSF is the lung shunt fraction.

The mean dose to the perfused normal liver compartment, DNLp, is calculated as:

  • DNLp [Gy] = Dtumor [Gy]/R.

Lung dosimetry

Regions of interest are drawn around the lung region and the liver region in both the anterior and the posterior views of planar scintigraphy. The view with the highest 99mTc-MAA counts for the liver region is used to determine the 99mTc-MAA counts in the lung region and the 99mTc-MAA counts in the liver region for the calculation of the lung shunt fraction (LSF). The LSF is calculated using the following formula:

  • LSF = (counts in lung region)/((counts in lung region) + (counts in liver region))

The mean lung dose (Dlung) corresponding to a specific LSF can be determined as:

  • Dlung [Gy] = Aadmin [GBq] × LSF × 49.7 [Gy-kg/GBq]/Mlung [kg], where Mlung is the mass of lung tissue [kg].

The Mlung can be calculated using a patient-specific CT scan or can be estimated as 1 kg.

Treatment plan dosimetry constraints

Normal liver constraints: patients are excluded if < 33% of the total liver volume based on body surface area (TLVBSA) is tumor free, and therefore, left untreated. The TLVBSA is calculated based on the method of Vauthey et al. [15] as:

  • TLVBSA [ml] = − 794.4 + 1267.3 × (height [cm] × weight [kg]/3600)0.5

If the 90Y treatment-spared liver volume is ≥ 33% but < 40% of the TLVBSA, the dose to the perfused normal liver parenchyma will be limited to < 150 Gy [14]. If the 90Y treatment-spared liver volume is ≥ 40% of the TLVBSA, there will be no upper limit to the dose allowed to the perfused normal liver parenchyma [18].

Lung mean dose (Dlung) constraints: Dlung ≤ 30 Gy for a single treatment, and the cumulative lung mean dose < 50 Gy if the patient is re-treated.

Tumor mean dose (Dtumor) constraints: the planned mean dose to the tumor must be in the range of 150–400 Gy. The lower limit of 150 Gy was based on the estimated mean dose required to achieve a complete response (CR) in previous studies [19, 20]. In the unlikely event that the planned mean dose to the total tumor compartment cannot exceed 150 Gy, the patient is considered a screening failure.

SIRT procedure

All treatments are performed with 3-day FLEXdose delivery option in a single treatment session using a selective or lobar approach. To optimize the predictive value of the 99mTc-MAA intra-hepatic distribution, the catheter tip during 99mTc-MAA injection is recommended to be placed at the same anatomical position from which 90Y resin microspheres will be administered (a deviation greater than ± 0.5 cm is allowed, if necessary, but must be documented; likewise, the occurrence of stasis will be documented).

After the SIRT procedure

Post-treatment assessment is conducted by imaging of the patient’s abdomen with 90Y SPECT/CT or 90Y PET/CT. Post-SIRT images are assessed by the Eligibility Review Committee using FDA-approved software to qualitatively confirm tumor targeting, quantify the mean tumor dose delivered, and to decide if re-treatment may be necessary.

Repeat SIRT

If the mean tumor dose delivered, based on post-treatment 90Y SPECT/CT or 90Y PET/CT imaging, was estimated to be < 100 Gy, SIRT re-treatment may be performed at the discretion of the treating physician within 4 weeks of the index procedure.

Assessments

Treated patients will be assessed at baseline (no more than 28 days before SIRT), and at 1, 2, 4, 6, 9 and 12 months (± 2 weeks) and at 24 months (± 1 month) after SIRT. All patients are assessed according to the schedule summarized in Table 2. All images used to assess tumor response are reviewed at an independent centralized laboratory (American College of Radiology Center for Research and Innovation, Philadelphia, PA; ACR). Patients may be withdrawn from the study due to loss of follow-up, death, withdrawal of consent, adverse events, investigator decision, or if there is failed SIRT implantation. If a patient withdraws from the study a study completion electronic case report form must be completed, describing the reason for discontinuation. After 24 months and exit from the study, patients will be assessed and managed as per usual care.

Table 2 DOORwaY90 trial assessment schedule

Outcomes

The co-primary endpoints are localized objective response rate (ORR; CR or partial response [PR]) using modified Response Evaluation Criteria in Solid Tumors (mRECIST) [21] with best response through 9 months, and duration of response (DoR), defined as the time-interval from first achieving a response (i.e., CR or PR) until disease progression (≥ 6 months for ≥ 60% of responders).

Secondary endpoints include: grade ≥ 3 toxicity (Common Terminology Criteria for Adverse Events, CTCAE v5.0) at 3 months and 6 months; quality of life (QoL) measured by Functional Assessment of Cancer Therapy-Hepatobiliary (FACT-Hep) questionnaire and the five-level EuroQol five-dimensional questionnaire (EQ-5D-5L); liver resection rate; and liver transplant rate.

Sample size calculation and statistical considerations

Continuous data will be summarized using descriptive statistics: sample size, mean, standard deviation, standard error, median, and range or interquartile range. Discrete variables will be summarized using frequency counts and percentages. All statistical analyses will be performed using SAS (version 9.4 or higher, SAS Institute Inc. Cary, NC), R (version 3.2 or higher, R Foundation for Statistical Computing, Vienna, Austria) or other widely accepted statistical or graphical software.

Primary analyses will be conducted under the principles of intent-to-treat (ITT), using the full analysis set (FAS) as defined in ICH E9 (Statistical Principles for Clinical Trials). Additionally, a per-protocol (PP) population is defined, including ITT patients who met all inclusion and exclusion criteria, received the intended dose of 150–400 Gy, and had no major protocol violations.

Methods for replacing missing data such as last value carried forward or multiple imputation will not be routinely used for the primary study analyses. For the purposes of defining the primary endpoints, death before the first attempt at evaluation will be defined as failure of the endpoint in question, and if no evaluable mRECIST data are collected for a patient, they will be classified as a failure for the ORR endpoint and as missing data for the DoR endpoint. Planned subgroup analyses will be conducted to evaluate consistency of results by gender and dosimetry method. Additionally, a dose–response analysis will be conducted to estimate the threshold mean tumor dose for response for tumors ≥ 3 cm. For the co-primary endpoint ORR, all lesions in the treated portion of the liver will be included for analysis, with up to 2 lesions per patient defined as target lesions and the rest non-target lesions. ORR results will be presented as the proportion of successes (CR or PR) with corresponding confidence limits calculated by the exact binomial (Clopper-Pearson) method. The performance goal is a lower confidence interval for best response on ORR of 40%. A DoR of ≥ 6 months in ≥ 60% of responding patients is the goal of the co-primary endpoint.

An overall summary of adverse events (AEs) will be presented, which will include the number and percentage of patients with at least one AE, and the total number of AEs. All AEs will be adjudicated by a Clinical Events Committee (CEC). AEs will additionally be classified according to the Medical Dictionary for Regulatory Activities (MedDRA) dictionary, summarized by severity, relationship to SIRT, outcome and seriousness. Serious adverse events (SAEs) will be summarized according to severity and relationship. Multiple occurrences of the same AE are counted once at the maximum grade and strongest relationship, as appropriate.

The postulated success rate for the ORR primary endpoint is estimated to be 55–60% based on previous experience with the study device and expert opinion on its effectiveness in the target population; as stated, the performance goal applied to the lower confidence bound for this estimate is 40%. Given a two-sided test at an alpha of 0.05 and desired power of 80%, the required evaluable sample size using the exact method for a single proportion and the worst estimate above is no greater than 90. Therefore, a study sample size of 100 is targeted to provide adequate powering for hypothesis testing as described above. Consent from up to 150 patients is anticipated to achieve the sample size of 100 patients, based on previous experience of eligibility failure of 33%. The overall study duration, from screening the first patient to the final follow-up visit, data analysis and final report, is expected to be approximately 39 months.

Discussion

The DOORwaY90 trial will provide clinically relevant and novel information on the effectiveness and safety of 90Y resin microspheres for the treatment of unresectable HCC when personalized dosimetry methods are used. Previous large trials of SIRT with 90Y resin microspheres have not used personalized dosimetry [11,12,13], and subsequent analyses have suggested that sub-optimal doses have been used in these trials, resulting in sub-optimal effects of SIRT [19, 22]. A recent trial using personalized dosimetry with 90Y glass microspheres further emphasized the need for a personalized approach to SIRT dosimetry to achieve optimum results [23]. However, because of differences in the size, distribution, and specific activity of glass and resin microspheres, dosimetry findings cannot be translated from one product to the other, and therefore, the DOORwaY90 trial is needed to provide information on the impact of personalized dosimetry on the efficacy of 90Y resin microspheres.

The delivered mean tumor dose of < 100 Gy to trigger possible SIRT re-treatment is 66% of the minimum target tumor dose. This cut-off was used partly because there is an error of at least 20% in quantifying post-SIRT 90Y SPECT/CT or 90Y PET/CT-based tumor doses [24]. Furthermore, an average tumor dose of 110 Gy has been shown to result in tumor response in > 50% of patients (according to RECIST; and according to EASL guidelines, response was achieved in > 70% of patients) [20]. Therefore, the mean tumor dose cut-off of 100 Gy should avoid unnecessary SIRT re-treatment.

There is increasing evidence that SIRT may have the potential to downstage unresectable HCC, or act as a bridge to transplantation. This needs investigation in dedicated clinical studies, but the DOORwaY90 trial includes patients with earlier stage HCC than in previous prospective trials of 90Y resin microspheres and may help provide further insight into the ability of SIRT to convert patients to resectability. The DOORwaY90 trial further offers the opportunity to assess the safety profile of 90Y resin microspheres in the US population—an important consideration as many systemic treatments for HCC have safety profiles that can adversely impact patient QoL.

The DOORwaY90 study design is unique for several reasons. The DOORwaY90 study is the first prospective, multicenter US trial to utilize personalized 90Y dosimetry with 90Y resin microspheres: the use of a patient-specific target TNR allows a personalized prescribed activity that results in a dose that achieves maximum damage to the tumor, with minimal damage to healthy tissue. It is also the first study to employ an Eligibility Review Committee to determine patients’ treatment plans. In addition, unlike previous studies, tumor responses will be evaluated by an independent review body (ACR), and using novel imaging software (MIM Sureplan) we will assess if tumors achieve a minimum threshold dose to determine if patients are eligible for early retreatment. This study also utilizes a well-controlled patient population with a well-defined safety margin. The FLEXdose delivery program for 90Y resin microspheres ensures that all sites are using the same activity per sphere, that increases confidence in the entire planned activity being delivered without the potential for unplanned stasis (the 3-day FLEXdose delivery option also reduces further any risk of a macroembolic effect). The co-primary endpoints of ORR and DoR used in the DOORwaY90 trial were chosen as a measure of local disease control. As SIRT is a locoregional therapy, assessment of local disease control is a more relevant measure of effectiveness than overall survival.

The results of the DOORwaY90 trial will help refine the potential benefits of SIRT with 90Y resin microspheres as first-line therapy for HCC and may strengthen future consensus guidelines. If the DOORwaY90 trial meets its co-primary endpoints, personalized SIRT may become a new standard of care for patients with unresectable HCC. Such a treatment has the potential to facilitate liver resection or transplantation, and a consequent increased life expectancy and enhanced QoL for these patients.