Colorectal cancer frequently spreads to the liver, a condition called colorectal liver metastasis (CLM). When surgery to remove these liver tumors is not possible because of their location, size, or the patient's health, doctors need alternative treatments to control the disease.
Stereotactic body radiation therapy (SBRT) is a precise form of radiotherapy that delivers very high doses of radiation to a tumor in just a few sessions, while minimizing damage to surrounding healthy liver tissue. It has emerged as a valuable option for patients with CLM who cannot undergo surgery.
Radiation dosing in SBRT is described using a concept called the biologically effective dose (BED10), which accounts for how the radiation is split across sessions. Doctors prescribe a certain dose, but the actual amount delivered to every part of the tumor can vary. This study asked an important question: does the variation in the lowest dose received by any part of the tumor affect how well the treatment works?
Specifically, the researchers focused on "cold spots": areas within the radiation treatment zone that receive less radiation than intended, even when the overall prescription seems adequate. Understanding whether these cold spots drive treatment failure could change how radiation plans are designed.
The team conducted a retrospective study at Samsung Medical Center in Seoul, South Korea, reviewing records of patients treated between May 2016 and June 2024. A total of 116 patients with 128 liver metastases from colorectal cancer were included. All had received SBRT with a minimum biologically effective dose of 70 Gy BED10.
For each treated lesion, the researchers extracted detailed dose-volume histogram (DVH) measurements from the original treatment plans. These included the maximum dose (Dmax), minimum dose (Dmin), mean dose (Dmean), and the dose delivered to specific percentages of the tumor volume (D2%, D95%, D98%, D99%). The most important measure analyzed was PTV Dmin, the minimum dose delivered anywhere within the planning target volume.
The main outcome studied was freedom from local progression (FFLP), which measures how long the treated tumors remained controlled after treatment. Statistical models including Cox regression were used to determine which factors, both clinical and dosimetric, most strongly predicted whether a tumor would eventually grow back.
Two treatment technologies were used: conventional X-ray radiation (XRT) in 39 cases and proton beam therapy (PBT) in 89 cases. Most patients were classified as having oligometastatic disease, meaning five or fewer active metastatic tumors at the time of SBRT.
With a median follow-up period of 22.3 months, the researchers found that 74.5% of treated tumors remained controlled at one year, dropping to 58.8% at two years. These rates reflect the known radioresistance of colorectal metastases, which require higher radiation doses to achieve durable control compared to liver tumors from other cancers such as breast cancer.
Overall survival was 77.2% at one year and 54.3% at two years. Progression-free survival (meaning no tumor growth anywhere in the body) was 37.6% at one year and 16.6% at two years, reflecting the aggressive nature of metastatic colorectal cancer even when individual tumors are well controlled with radiation.
In univariable analysis, many factors were associated with whether a tumor recurred locally, including the extent of metastatic spread, number of prior chemotherapy lines, pre-treatment CEA levels, and multiple dosimetric measures. However, when tested together in a multivariable model, only two factors independently predicted tumor recurrence: polymetastatic disease (having more tumors at the time of treatment) and PTV Dmin (the minimum dose inside the treatment zone).
Patients with polymetastatic disease had more than three times the risk of local recurrence compared to those with oligometastatic disease. Every 10 Gy BED10 increase in the minimum dose reduced the risk of local recurrence by approximately 11%, highlighting how meaningful even small improvements in minimum dose delivery can be.
A striking finding in this study was that despite many patients receiving the same nominal prescription dose, their actual PTV Dmin values varied enormously. Two-dimensional frequency maps showed that cases with identical prescriptions could have very different minimum doses, confirming that the prescription dose alone does not fully describe what was actually delivered to every part of the tumor.
Using an optimal threshold of 100 Gy BED10 as the minimum dose cutoff, patients were divided into low and high PTV Dmin groups. Those in the low group (below 100 Gy BED10 minimum dose) had a one-year FFLP of only 64.2%, compared to 87.1% in the high group. This difference was highly statistically significant and clinically meaningful.
Tumor control probability (TCP) modeling further confirmed the superiority of PTV Dmin as a predictor. The TCP curve for PTV Dmin was significantly steeper than for prescription dose, meaning that small changes in minimum delivered dose translate into large differences in the probability of controlling the tumor. The model estimated that achieving 80% tumor control at one year required a minimum dose of approximately 100 Gy BED10.
These findings support the biological concept that underdosed regions, or cold spots, within the tumor may harbor more radioresistant or hypoxic cancer cells. If these cells receive insufficient radiation to be killed, they can regrow and become the source of local treatment failure.
Colorectal cancer cells are known to be relatively resistant to radiation compared to other tumor types. Prior research has shown that achieving durable local control requires delivering substantially higher radiation doses than for, say, lung or breast cancer metastases in the liver. Some studies have estimated that preventing tumor regrowth in 70% of cases requires doses exceeding 185 to 250 Gy BED10, which is extremely high and difficult to achieve safely.
The challenge is that the liver is a radiosensitive organ. Delivering very high doses risks causing radiation-induced liver disease (RILD), a serious complication where the liver loses function after treatment. In this study, RILD occurred in 14.1% of cases, and it was numerically more common after X-ray radiation (20.5%) than proton therapy (11.2%), although the difference was not statistically significant.
Proton beam therapy (PBT) was used in nearly 70% of cases in this study and offered a promising advantage: its physical properties allow radiation to stop precisely at the tumor without passing through and depositing dose in healthy tissue on the other side. This may allow slightly higher tumor doses while better protecting surrounding liver. The PBT group in this study received slightly higher BED10 with lower RILD rates, supporting proton therapy as a tool to safely escalate doses in CLM.
The findings underscore a critical clinical principle: in colorectal liver metastases, it is not enough to aim for the right average or maximum dose. Ensuring that every part of the tumor receives adequate dose, including the lowest-dose region, may be more important than achieving a high prescription dose on paper.
The key practical implication of this study is that radiation oncologists should consider adding a PTV Dmin constraint to SBRT planning for colorectal liver metastases. Rather than simply aiming for a high prescription dose, treatment plans should explicitly require that the minimum dose anywhere within the tumor exceeds a defined threshold, such as 100 Gy BED10.
Current SBRT planning typically focuses on ensuring that 95% of the tumor volume receives the prescription dose. But this means up to 5% of the tumor could receive substantially less. For colorectal metastases, that underdosed 5% may be enough to cause treatment failure, particularly if it contains radioresistant or poorly oxygenated cells.
The researchers also suggest that incorporating additional information such as radiomics (quantitative features extracted from CT or MRI scans) or genomic tumor characteristics could further refine patient selection and dose planning, potentially improving the model's ability to predict outcomes beyond the 0.673 C-index achieved in this study.
The authors are transparent about limitations: this was a retrospective single-institution study with heterogeneous treatment protocols. The tumor control probability curves have wide confidence intervals, limiting their direct use as rigid prescriptive thresholds. Prospective clinical trials with standardized PTV Dmin-based planning objectives are needed to formally validate and implement these findings.
This study establishes that the minimum dose delivered within the tumor, specifically the PTV Dmin, is the single strongest predictor of whether colorectal liver metastases will be controlled after SBRT. It outperforms the prescription dose and all other dosimetric measures tested.
For patients with colorectal liver metastases being considered for SBRT, this research suggests that how the radiation is distributed throughout the tumor matters enormously. A perfectly prescribed dose that still leaves cold spots inside the tumor may result in cancer growing back from those underdosed areas.
The finding that polymetastatic disease also strongly predicted worse outcomes is equally important. Patients with more active tumors at the time of treatment have higher rates of local recurrence even after technically adequate radiation. This may reflect the more aggressive biology driving widespread metastatic spread.
Going forward, incorporating PTV Dmin as a standard planning goal alongside better tools such as proton therapy to safely escalate minimum doses could improve local control rates for colorectal liver metastases, offering more patients a chance at lasting disease control from this difficult-to-treat scenario.