Effects of gold fiducial marker implantation on tumor control and toxicity in external beam radiotherapy of prostate cancer

Prostate cancer can be treated by external beam radiotherapy (EBRT) brachyradiotherapy or radical prostatectomy; for low-risk disease and highly selected intermediate disease, active surveillance can be offered.^{1, 2, 3, 4, 5} Important factors in patient decision making are, therefore, the duration of treatment, circumstances of treatment, and treatment-induced side effects. Regarding side effects in EBRT, the use of intensity-modulated radiotherapy (IMRT) has been shown to reduce treatment-associated side effects compared with 3D-conformal radiotherapy.^6,7 Further reduction of side effects is expected using image-guided radiotherapy (IGRT)⁸, in which fiducials, ultrasound, MRI, CT scans, or electromagnetic responders are used to locate the prostate before and during radiotherapy.⁹ However, to our knowledge, there are as yet no data regarding IGRT in the form of randomized controlled trials, although IGRT is recommended by the NCCN¹, EAU⁵, British NICE¹⁰, and German S3⁴ guidelines.

With this paper, we explore the role of gold fiducial markers (GFM) used for IGRT in primary prostate cancer treatment. There are only a few studies comparing patients with and without GFM. For example, a retrospective study by Zelefsky et al.¹¹ described an advantage after implanting GFM in regard to genitourinary (GU) side effects in all patients and biochemical no evidence of disease (bNED) rates in high-risk prostate cancer patients. However, this study compared a group of patients treated with IGRT, but without IMRT with an IMRT-only group, making its results difficult to apply to modern day radiotherapy, in which IMRT is always recommended.¹ The same applies to Sveistrup et al.¹², Zapatero et al.¹³ and Wortel et al.⁷, which compared patients treated with fiducial-marker IGRT and IMRT with patients treated with 3D conformal radiotherapy.¹² Singh et al. compared the use of GFM IGRT with no IGRT et al.¹⁴ However, to our knowledge, only one study has compared the use of IGRT with and without GFM¹⁵, and it showed no significant benefits regarding tumor control and toxicity.

In our department, we use cone-beam CT scans and ExacTrac (Brainlab, Munich, Germany), as well as routinely implanted GFM for image guidance. As GFM are implanted with the use of anesthesia, the benefit of the GFM has to be larger than the risks from anesthesia, including, but not limited to, nausea, allergic reactions, intraoperative awareness, or death^16,17, and the risk of the intervention itself, such as infections or injuries to adjacent organs. Local anaesthesia reduces the risks of anaesthesia. Some patients have contraindications against anesthesia or refuse the intervention. A more precise patient positioning due to GFM can be expected to lead to greater treatment precision and hence, possibly, to reduced toxicity and increased tumor control. These assumed benefits can justify its routine application in patients with prostate cancer. Without these advantages, the routine use of GFM would not be justified. Therefore, we wanted to compare a group of patients receiving IGRT with GFM with a group receiving only IGRT to evaluate the potential benefits of GFM regarding tumor control and GI and GU toxicity.

Patient characteristics are displayed in Table 1. We also analyzed the reasons for patients not receiving GFM implantation. Among them, most patients (12, 60%) had medical contraindications for anesthesia, 5 patients (25%) refused GFM implantation, and for 3 patients (15%) who did not receive GFM, the reason was unknown.

bNED rates are displayed in Figure 1. bNED ratios after 5 years for patients with and without GFM implantation were 92% and 100%, respective-

Figure 1

ly. After 8 years, bNED rates were 87% and 86%, respectively (p = 0.61). We also analyzed disease-specific survival (DSS) and overall survival (OS). Regarding DSS, we did not find a single death due to prostate cancer (p > 0.9999). For OS, survival proportions for patients with and without GFM after 5 years were 93% and 95%, and after 8 years 90% and 95% (p = 0.56).

Maximum acute and late gastrointestinal side effects are displayed in Tables 2 and 3. No significant differences between the two groups were detected. It is noteworthy that at no point was a grade 4 toxicity detected. We also compared late GI and GU toxicity in the two groups after 3, 12, 24, 36, 48, 60, 72, 84, and 96 months, and did not find a significant difference at any point.

For DVH data, we started by comparing the two groups regarding volumes, Dmax/D1, and Dmean of the CTV, PTV, rectum, and bladder, as well as the Dmin of the CTV and the V70 and V60 for the rectum and V70, V50, and V30 for the bladder. Significant differences were found in the PTV (median 124.7 cm³with GFM and 157.1 cm³ without, p = 0.004), the Dmin of the CTV (median 76.0 Gy with GFM and 76.5 Gy without, p = 0.04), the rectal Dmean (median 36.0 Gy with GFM and 39.3 Gy without, p = 0.01), V70 (median 11.4 Gy with GFM and 16.1 Gy without, p < 0.001), and V60 (median 20.0 Gy with GFM and 25.3 Gy without, p < 0.001). An overview of the DVH data is presented in Table 4.

We also performed a univariable, and if more than one variable was significant, a multivariable Cox regression analysis regarding bNED and OS, as displayed in Table 5, as well as the onset of late GI or GU toxicity grade 2 or higher. The results of our Cox regression analysis regarding bNED and OS are displayed in Table 5.

As no patient died of prostate cancer, we did not model a Cox regression for DSS. Analyses including rectal or bladder DVH variables regarding bNED and OS were performed, and none of the variables was significant. Therefore, to improve clarity, we did not add them to Table 5. The results regarding toxicity are displayed in Table 6.

As GFM implantation is an invasive procedure, the benefits of implantation have to outweigh the potential risks. There are two categories of risks. The first is the risk due to the implantation itself. Citing our information sheet, these risks are bleeding, injuries of the bladder and urethra, infection including abscess formation, and allergic reactions to the prescribed antibiotic. The second category is risks due to the anaesthesia. Even when only a breathing mask is used to administer anaesthesia, these risks include tissue damage due to patient positioning, allergic reactions, malignant hyperthermia, subsequent confusion, aspiration, and regaining consciousness during anesthesia.^16,17 While these side effects occur rarely in our clinical experience, as well as in the literature²³, they have to be considered in evaluating GFM implantation, aside from its influence on tumor control and side effects.

Several retrospective studies comparing fiducial IGRT with a control group exist. However, most of them compare a 3D conformal group with either an IGRT 3D conformal group^14,24 or an IGRT IMRT group¹², making them hard to apply to today’s treatment due to their being outdated or their comparison of two different treatment modalities with inherent differences regarding outcomes. While many international guidelines^1,4,10 suggest the use of IGRT, no study covering this topic is mentioned in the NCCN¹ or NICE¹⁰ guidelines. Napieralska et al.¹⁵, to our knowledge the only other study comparing patients with and without fiducial marker implantation and IMRT, found no significant differences regarding either bNED, with the exception of improved OS in intermediate patients, or late toxicity when comparing fiducial marker guidance and bone structure guidance. A significant difference was found regarding acute GU toxicity. Our results are similar, with no significant differences in acute toxicity. However, the aforementioned study did not include DVH data. Looking at our DVH data, we were unable to translate the differences regarding PTV and rectal variables into differences related to tumor control or toxicity, most likely due to DVH constraints in both groups.

It is quite likely that the most crucial effect regarding potential outcomes in our department is the safety margin reduction by 3 mm in GFM patients, which is absent from the aforementioned studies. The effect of reducing the safety margin by 3 mm can be displayed using math. Assume a spherical form for the prostate, with a radius between 2 and 2.5 cm and safety margins of 7 mm and 10 mm. The radii, including the PTV safety margins, are 2.7 cm and 3 cm, with an initial radius of 2 cm and 3.2 cm, respectively, and 3.5 cm with an initial radius of 2.5 cm. With 4/3*π*r³ being the formula for the volume of a sphere, reducing the safety margins by 3 mm leads to a decrease in the PTV of 27% for the 2-cm initial radius and 24% for the 2.5-cm initial radius (2.7³/3³and 3.2³/3.5³). In our data, the difference in the median volume is 21%, similar to the expected difference. Although this calculation is far from perfect, as the prostate is not a perfect sphere, it allows one to imagine the effect of safety margin variation through the third power of the radius. Zelefsky’s conclusion in his IGRT study¹¹ was that safety margins should be reduced. However, to our knowledge, he has yet to publish any data comparing bNED rates before and after margin reduction.

When comparing our bNED rates with those of groundbreaking studies like those of Peeters et al.²⁵ and Pasalic et al.²⁶, using 78 Gy, and Dearnaley et al.²⁷, using 74 Gy, with bNED rates after 5 years ranging from 64% for Peeters to above 90% for Pasalic, we are leaning toward the top end, with bNED rates after 5 years of 92% and more, while only including patients with low- and intermediate-risk tumors. Regarding late side effects, while there were no significant differences between the two groups, our data, with 29%–35% of maximum GI toxicity at RTOG grade 2 or higher toxicity and approximately 36%–40% of GU toxicity of RTOG grade ≥ 2, are worse than the results presented by Napieralska¹⁵, showing 12% grade ≥ 2 GU side effects and 15%–19% grade ≥ 2 GI side effects, possibly due to differences in side-effect assessment between institutions. Zelefsky reports lower toxicity rates, using CTCAE criteria, complicating a direct comparison.

One weakness of our study is its retrospective nature. On top of that, there is no group of patients with GFM implantation and a safety margin of 10

mm, complicating the direct comparison of our two groups, as GFM were implanted and the safety margin was reduced to 7 mm together. However, retrospective studies should be used to generate hypotheses, not prove them, and with this study we have generated the hypothesis that the use of GFM for IGRT provides no benefits, but only risks, and should not be performed regularly if other, non-invasive tools for IGRT are in use. Another weakness is the limited sample size.

Regarding strengths, we analyzed a homogeneous patient collective to address this question, with in-depth data including bNED, OS, DSS, and acute and late toxicity, as well as DVH data. Besides, our patient collectives were recruited in parallel, and this reduced any potential bias due to changing treatment modalities over time.

With this study, we have developed the hypothesis that the use of GFM in IGRT does not provide a substantial benefit regarding tumor control or toxicity when other modalities for IGRT are used. To clarify the role of GFM in IGRT, prospective studies based on this hypothesis are needed to possibly reduce the number of unnecessary medical interventions in the treatment of men’s most common cancer.