The answer is both yes and no. Yes, in that we are both part of the same team, the MD Anderson Section of Outreach Physics, which comprises IROC-Houston, the RDS, and the Accredited Dosimetry Calibration Laboratory (ADCL). Our collective mission in Outreach Physics is to broadly reach out to the radiation oncology community in the United States and worldwide to ensure patient safety through independent quality assurance (QA) services. Both IROC and the RDS provide independent peer review of beam output calibrations for external beam radiotherapy machines through a mail audit system. Both IROC and the RDS are located at the same physical address (with our labs/offices occupying different portions of the building) and we have a very collaborative relationship.

The key difference between the two groups is to whom we offer services. IROC provides output checks for institutions that participate in clinical trials. The output checks are a requirement of participation. Specifically, IROC monitors all photon beams and three electron beams for each linac at participating institutions. The RDS is a voluntary for-fee service that is available on-demand. Many RDS customers participate in clinical trials and have annual beam monitoring through IROC. For those dual-service institutions, the RDS monitors the electron beams not monitored by IROC, with shipment times that may be coordinated and aligned. Some of the IROC centers also use the RDS for more frequent (e.g., mid-year or quarterly) output checks.

Another difference between IROC and RDS is that we use different dosimeters in our mail audit systems. IROC uses optically stimulated luminescent dosimeters (OSLD) and RDS uses thermoluminescent dosimeters (TLD). Both dosimeters have similar uncertainties of <1.5% as described in AAPM Task Group Report 191 (Kry et al 2019). Also, the doses reported by both IROC and RDS are traceable to a primary standard and based on TLD/OSLD irradiated at a Secondary Standards Dosimetry Laboratory, the MD Anderson ADCL.

What we mean by on-demand is that our customers choose (1) which beams they want to monitor, e.g., all beams, some beams, or a single beam; (2) at what frequency they want service, annual, semi-annual, quarterly, monthly, or by request, which is common after a new linac is commissioned or after a major repair; and (3) shipment month(s). Our annual customers typically align their output checks with their annual TG-51 calibration.

In general, more frequent monitoring provides extra peace of mind in knowing that if an output issue were to arise, there would be a shorter time interval between checks, decreasing the length of time between when the issue(s) first arose and when it was detected. Also, data from the Global Harmonization Group (GHG; Kry et al 2018), which comprises multiple international independent peer review groups, demonstrate the substantial value of routine and/or periodic auditing of all radiotherapy beams. In a recent survey of 210,167 worldwide audit results, the GHG reported that (1) machine calibrations are challenging owing to the number of processes involved that can go awry; (2) calibration errors originate from a wide range of sources; (3) calibration errors can impact a single beam, multiple beams, or all beams at a facility; and (4) calibration errors can be introduced into beams with historically accurate output.

We order from an Ohio-based company, Quantaflux LLC (Dayton, OH). When our TLD capsules are running low, we contact Quantaflux and they begin the process of growing a large batch lithium fluoride crystal for us. Depending on the size of our order, it takes several months for a crystal batch to be sufficiently large. Once harvested, the batch of crystals is ground down to a fine powder, which is thoroughly mixed, annealed, and finally loaded into small capsules. We typically order between 100,000 and 200,000 TLD capsules, which last between two and three years. Each new batch undergoes a complete batch commissioning (as described in TG-191), which includes characterizing energy correction, linearity, and fading factors. Once our internal QA process is complete and we’ve established reproducibility, we begin disbursement. The RDS names our TLD batches for the year in which they were annealed; we are currently on Batch-19.

We irradiate standards and controls every week at the ADCL. These are read at the beginning, middle, and end of each TLD reading session. We periodically evaluate the batch for consistency and reproducibility. We also do periodic intercomparisons with IROC. Additionally, we participate in an annual International Atomic Energy Association (IAEA) “blind check” and reference irradiations. In these, RDS sends the IAEA five sets of dosimeters for irradiation with a Co-60 beam at the IAEA laboratory at doses ranging from 2.5 to 3.5 Gy, with a step size of 0.25 Gy, and one set for the blind check. They return the irradiated TLDs and we analyze the doses as we would for any other center. We just completed the 2019 IAEA audit and our results were within 1% agreement with the stated IAEA doses.

We irradiate standards and controls every week at the ADCL. These capsules are read at the beginning, middle, and end of each TLD reading session. We periodically evaluate the batch for consistency and reproducibility. We also do periodic intercomparisons with IROC for both photon and electron energies. Additionally, we participate in an annual International Atomic Energy Association (IAEA) “blind check” and reference irradiations. The 2019 audit results were within 1% agreement with the stated IAEA doses.

Our first priority is to use a passive detector that is easily mailable with low uncertainty. Because both TLD and OSLD meet these criteria and have very similar uncertainty, our decision to continue to use TLD was based on cost and experience. First, from a cost perspective, the initial ramp-up to switch from TLD to OSLD would be very expensive because it would require replacing our TLD readers (we have six), re-training our staff to use a new type of reader, and fabricating new blocks for a different dosimeter. These changeover costs would ultimately have to be passed on to our customers. Since it is important to us to maintain a low-cost service, making it affordable for any institution, this is not ideal. Also, the RDS has a more than 30-year history using TLD. The system works, it’s cost effective, and the uncertainty is both low and well characterized. So for now and the foreseeable future, we are staying the course with TLD.

We are part of MD Anderson, but we are a non-profit self-funded service, meaning our fees go directly to running our service, including staff salaries, TLD capsules, TLD readers and their maintenance, other equipment such as ion chambers and electrometers, office supplies, etc. We also share in some of the operational costs for the units owned by the ADCL where we irradiate our TLD standards and controls.

The AAPM Task Group 103 recommends independent output checks for all beams on a routine basis. Similarly, the American College of Radiology (ACR) requires monitoring all beams on an annual basis to receive and maintain ACR Radiation Oncology Accreditation. However, beyond the recommendations and requirements, it just makes sense from a patient safety perspective. Although calibration errors are rare (identified in approximately 1% of beams monitored), they can lead to very detrimental harm for patients. Furthermore, such errors can be financially devastating to the institutions in which they occur, especially in the context of the cost of peer review. It costs <$1000 for output checks for all beams for a typical linac (two photons and five electrons), which translates to about $4 to $5 per new start on a typical volume linac (with 200–250 new starts). So, to an administrator, I’d say “independent review makes sense first and foremost for patient safety, aligns our center with national and international recommendations, demonstrates that our center is serious about QA, and provides a low-cost safety net.”

Most of our output checks show good agreement between doses measured by RDS and doses reported by the institution (RDS/institution ratio); 98.6% of all beams agree within 5% and 90.9% agree within 3%. But, that also means that just over 1% of the beams that we monitor are outside of the 5% agreement criterion. One example of a calibration issue that always comes to mind was a check for a newly commissioned linac that was ready to “go live.” The physicist sent us a rush order for TLDs for all beams. When we analyzed the results, we reported dose differences between 6% and 7% from those stated by the institution. We reached out to the physicist, who repeated the calibration but found nothing different from initial calibration. To be safe, he again repeated the measurements with a different ion chamber and electrometer. Again, he observed nothing out of order. Then, while he was taking down the equipment, the electrometer cable fell apart. He repeated measurements one more time using a new cable. This time, he observed the doses to be 6% to 7% lower than during his previous measurements. He adjusted the beam output and ordered additional TLDs from us. This time the institution’s stated doses and the RDS measured doses were in agreement for all beams. Of particular importance in this story is that the calibration error was caught before a single patient was ever treated. Imagine what could have happened if they had not had independent peer review.

This really boils down to uncertainty. For our well-established TLD dosimetry protocol (three TLD loaded into each phantom), the standard error is 1.3% (Kirby et al 1992). Thus, a TLD check is considered acceptable if dose agreement is within 5% of the stated dose, corresponding to a 93% confidence limit. So, when a beam is outside of 5% agreement, we are confident that there is a real issue. Each year, the RDS provides reports for more than 12,000 beams, of which more than 150 do not meet the 5% agreement criterion.