Optimum 3D dose prescribing for prostate radiation therapy

Knowledge Exchange for Patients and Health Professionals

The goal of prostate radiation therapy

The goal of radiotherapy when treating patients with localised prostate cancer, is to distribute sufficient radiation dose to eradicate the tumour and as little as possible to all parts of the surrounding healthy tissues. Clinicians want to make sure that enough radiation dose reaches the prostate cancer to kill it, while keeping any radiation side effects to the surrounding healthy organ structures (bladder, urethra and rectum) to a minimum.

Optimising the dose distribution to kill all the cancer in the prostate with the least side effects is of paramount importance. It would provide better quality treatment for prostate cancer patients receiving radiation therapy. 3D analysis, of patient data taken in previous clinical trials, has not previously been published in the literature.

Treatment side effects

As already mentioned, the bladder, urethra and rectum are critical structures located very close to the prostate (Figure 1) and should receive as little X-ray dose as possible. As can be seen in Figure 2, delivering some radiation dose to these critical structures is inevitable as the radiation beams usually pass through them. But, due to the multi-directional X-ray beams used to treat the prostate and the close vicinity of the critical structures, it’s inevitable that the nearby normal structures will receive some fraction of the radiation dose intended to treat the prostate.

Side effects that may occur from the radiation therapy are:

• Haematuria: bladder damage causing bleeding that shows up in urine.
• Dysuria: a burning sensation in the urethra during urination.
• Tenesmus: a cramping pain in the rectum while passing stools.

3D Prescribing the treatment

Modern computer treatment planning software can accurately calculate the 3D dose distribution according to the radiation oncologist’s prescribed treatment protocol, The linear accelerators can then be used to precisely deliver the 3D dose distribution to the prostate tumour volume (Figure 2) . 

To ensure the best treatment outcome of optimum prostate cancer kill and minimum side effects to the critical bladder, urethra and rectum, the radiation oncologist needs to have learnt from clinical trials what X-ray dose is best to prescribe as:

• distribution to the tumour; and
• maximum dose limits to avoid side effects to any parts of the critical structures. 

Further valuable information (not previously established) would be:

• are there regions of the prostate to irradiate where the X-ray dose may be more crucial?

and

• which regions of the critical organ structures may be more sensitive to the X-ray dose received?

These were some of the questions that my research work sought to answer [1-3]. The results would help achieve better cure rates and fewer side effects for prostate cancer patients receiving radiation therapy.

Optimising the radiation dose prescription and 3D plan

Clinical studies have previously involved the collection of prostate radiation therapy patient data from cohort centres. The data is analysed after dividing it into groups treated with different radiation dose protocols. Researchers can then investigate whether the cure rates for patients treated with higher tumour dose are any better than patients with a lower prescribed dose.

By repeating this process over multiple studies, researchers determine the optimum dose to irradicate the tumour. Researchers could also investigate and establish an optimum dose limit the critical organs could receive to avoid side effects.

But this research method could only provide dose information describing the prostate tumour or critical organ as a whole. The clinical research trials could not provide a more detailed geographic result encompassing the 3D distribution of dose in either the prostate or the bladder, urethra and rectum.

My thesis research

To provide a better prostate cure outcome and limit critical organ detrimental effects, my thesis involved developing a method that could investigate the 3D dose distribution to the prostate tumour and critical organs (bladder, urethra and rectum). The results could assist the radiation oncologist when prescribing the patient’s prostate treatment plan.

A statistical mathematical model was developed to search point-by-point throughout the whole pelvic anatomy and determine where higher or lower dose points were associated with better or worse patient outcome. The 3D search point-by-point method aimed to identify:

• 3D tumour regions that were particularly crucial to cover with sufficient X-ray dose to stop cancer recurrence after treatment; and
• critical structures where there were found to be particularly radiation sensitive regions that should be avoided to prevent side effects.    

Method

Preparation of 3D image-dose matrix data

For the analysis, 3D dose distributions were calculated for a large cohort of prostate radiation therapy patients. Multiple slices showing the patient’s CT scan data (Figure 2(a) were superimposed with the dose calculation (Figure 2(b)). However, Figure 2 shows only a single slice – not a 3D dose distribution.

To create the 3D dose distribution, the image-dose data for multiple slices (as shown in Figure 3, left), are reproduced as a 3D matrix image with dose and anatomical information stored in the data file (Figure 3, right).

Deformable registration of patient data

Each of 680 patient 3D dose distributions were merged to one unique anatomical CT image template by a mathematical process called deformable registration.

Deformable registration allows each individual dose distribution to be ‘deformed’ (by computer software) to fit the same standard CT image (Figure 4). This is a crucial technique to apply because every patient’s anatomy is different. So, if we want to compare the patient results for one against each of the other 3D anatomical results, then we must ensure they are all aligned to the same template.

3D Search for dose-outcome relationships

For each patient in the cohort trial, outcome data (such as side effects and tumour recurrence) were stored with the registered 3D dose distributions. Statistical models were developed to search point-by-point through the pelvic anatomy data files to calculate the relationships between treatment outcome and radiation dose.

The models are complex. But it will be attempted here to describe the basic concept for how the search algorithm works.

For each point in the pelvic anatomy, we divide the cohort into a high dose group and a low dose group, based on the median dose (Figure 5.):

High-dose group: cohort patients with an anatomical point dose greater than the median dose.

Low-dose group: cohort patients with an anatomical point dose less than the median dose.

Then, we use a statistical model to compare the incidence of a particular patient outcome in the high dose group and the low dose group. That is, we want to know which dose group experienced a higher incidence of the outcome, and whether the difference between the two groups is statistically significant.

Repeating this process for every point in the pelvic anatomy can reveal clusters of points showing specific dose-outcome patterns.

Examples of negative outcome

Rectal bleeding: if the test for a particular anatomical point revealed that patients in the high dose group experienced more rectal bleeding than patients in the low dose group, then it can be concluded that the patient’s rectal anatomy at this point may be sensitive to radiation dose.

Any increase in dose to this point may result in more rectal bleeding and should be avoided.

Tumour kill: if the low dose group contains more patients with a tumour recurrence at a specific prostate point, then this point should have received greater dose for cancer cure.

Prostate clusters: When a cluster corresponds to a particular region in the prostate tumour site, then it indicates that the tumour should receive more X-ray dose in that cluster’s location to reduce the risk of cancer recurrence.

Critical organ clusters: When clusters occur in the critical organs, then this is an indication of radiation sensitive regions and the X-ray dose should be less to avoid the side-affects. The process for this analysis is illustrated in Figure 5.

Results

Finding dose-outcome relationships in 3D

The analysis found several pelvic anatomy dose-outcome associations. Three major associations will be presented here:

1. Patients who receive less than the median radiation dose at the posterior (rear end) of the prostate were more likely to have cancer recurrence (Figure 6a).

• The analysis indicated that without adequate dose coverage at the posterior of the prostate, cancer recurrence is more likely.

Consequently, it’s recommended that clinicians ensure adequate dose coverage of the posterior prostate region.

2. For patients who receive a radiation dose greater than the median dose at the distal urethra, the patient is more likely to experience dysuria in the urethra (Figure 6b).

• Urine from the bladder is excreted via the urethra and passes through the centre of the prostate receiving the X-ray treatment. Since the urethra receives the high prostate dose, the radiation is likely to damage the urethra tube wall and cause dysuria (a burning sensation felt during urination).

The data analysis highlighted radiation sensitivity at the distal urethra end.

3. Patients who receive greater than the median radiation dose posteriorly to the rectum are more likely to experience tenesmus (Figure 6c).

• The space posterior to the rectum contains a fat region which has previously been shown to be associated with tenesmus [4].

The analysis confirms this association, highlighting the need to avoid dose in this region to minimise the chance of tenesmus. 

Conclusion and Future Work

The recommendations listed above can assist the radiation oncologist when defining specific regions of dose to the prostate and critical structures. This helps to maximise better patient outcomes. This work indicates the need to go beyond prescribing dose constraints to the dose-volume histogram of the entire organ, but rather prescribing to sub-regions of the organ.

However, this work was very much an initial exploration using this analysis technique. The future direction should be to repeat the analysis on larger cohorts, and other patient populations, to reinforce these dose-outcome associations. It will also be possible to build models that incorporate other patient data too (like the shape and size of the tumour). These models could then produce custom dose distributions for any given patient that would maximise the chance of killing the tumour and minimise the chance of side effects.

References

1. Marcello, M., Denham, J.W., Kennedy, A., Haworth, A., Steigler, A., Greer, P.B., Holloway, L.C., Dowling, J.A., Jameson, M.G., Roach, D. and Joseph, D.J., 2020. Reduced dose posterior to prostate correlates with increased PSA progression in voxel-based analysis of 3 randomized phase 3 Trials. International Journal of Radiation Oncology* Biology* Physics, 108(5), pp.1304-1318.
2. Marcello, M., Denham, J.W., Kennedy, A., Haworth, A., Steigler, A., Greer, P.B., Holloway, L.C., Dowling, J.A., Jameson, M.G., Roach, D. and Joseph, D.J., 2020. Relationships between rectal and perirectal doses and rectal bleeding or tenesmus in pooled voxel-based analysis of 3 randomised phase III trials. Radiotherapy and Oncology, 150, pp.281-292.
3. Marcello, M., Denham, J.W., Kennedy, A., Haworth, A., Steigler, A., Greer, P.B., Holloway, L.C., Dowling, J.A., Jameson, M.G., Roach, D. and Joseph, D.J., 2020. Increased dose to organs in urinary tract associates with measures of genitourinary toxicity in pooled voxel-based analysis of 3 randomized phase III trials. Frontiers in oncology, 10, p.1174.
4. Gulliford, S.L., Ghose, S., Ebert, M.A., Kennedy, A., Dowling, J., Mitra, J., Joseph, D.J. and Denham, J.W., 2017. Radiotherapy dose-distribution to the perirectal fat space (PRS) is related to gastrointestinal control-related complications. Clinical and translational radiation oncology, 7, pp.62-70.

Marco Marcello PhD, 28 January 2022