Is intraoperative nomogram-based overplanning of prostate implants necessary?

A machine-learning approach based on 409 treatments to predict optimal number of iodine-125 seeds in low-dose-rate prostate brachytherapy

Purpose

Low-dose-rate brachytherapy is a key treatment for low-risk or favorable intermediate-risk prostate cancer. The number of radioactive seeds inserted during the procedure depends on prostate volume, and is not easy to predict without pre-planning. Consequently, a large number of unused seeds may be left after treatment. The objective of the present study was to predict the exact number of seeds for future patients using machine learning and a database of 409 treatments.

Material and methods

Database consisted of 18 dosimetric and efficiency parameters for each of 409 cases. Nine predictive algorithms based on machine-learning were compared in this database, which was divided into training group (80%) and test group (20%). Ten-fold cross-validation was applied to obtain robust statistics. The best algorithm was then used to build an abacus able to predict number of implanted seeds from expected prostate volume only. As an evaluation, the abacus was also applied on an independent series of 38 consecutive patients.

Results

The best coefficients of determination R² were given by support vector regression, with values attaining 0.928, 0.948, and 0.968 for training set, test set, and whole set, respectively. In terms of predicted seeds in test group, mean square error, median absolute error, mean absolute error, and maximum error were 2.55, 0.92, 1.21, and 7.29, respectively. The use of obtained abacus in 38 additional patients resulted in saving of 493 seeds (393 vs. 886 remaining seeds).

Conclusions

Machine-learning-based abacus proposed in this study aims at estimating the necessary number of seeds for future patients according to past experience. This new abacus, based on 409 treatments and successfully tested in 38 new patients, is a good alternative to non-specific recommendations.

A simple technique for the generation of institution-specific nomograms for permanent prostate cancer brachytherapy

Purpose

Nomograms once had a vital role in prostate brachytherapy practice. Although some of their functions have been assumed by computerized dosimetry, many programs still find them useful to determine the number and strength of seeds to be ordered in advance of the implant. As it has been demonstrated that brachytherapists differ in their implant practices and preferences (in regard to seed distribution and total implanted activity), we propose a simple technique for generating institution-specific nomograms.

Material and methods

Using the data generated by previous implants, we demonstrate a simple technique, utilizing ubiquitous software, for generating nomograms to predict seed number, strength, and total implant activity.

Results

Despite wide variations in the degree of post-implant swelling (+12% to +56%), nine of the first ten implants using the generated nomogram have met all target dose parameters.

Conclusions

It is appropriate for each institution to develop nomograms specific to their prostate brachytherapy technique and constraints. Nomograms can estimate the number and strength of seeds to be ordered, and may partially compensate for the effects of implant-induced swelling.

Optimal equations for describing the relationship between prostate volume, number of sources, and total activity in permanent prostate brachytherapy

OBJECTIVES

To determine whether there is an optimal type of mathematical equation for predicting seed and activity requirements for permanent prostate brachytherapy.

METHODS

Four institutions with extensive brachytherapy experience each submitted details of more than 40 implants. The data was used to generate power and linear equations to reflect the relationship between preimplant volume and the number of seeds implanted, and preimplant volume and the total implant activity. We compared the R and standard error of the generated equations to determine which type of equation better fit the data.

RESULTS

For the limited range of prostate volumes commonly implanted (20-60 mL), power and linear equations predict seed and activity requirements comparably well.

CONCLUSIONS

Linear and power equations are equally suitable for generating institution-specific nomograms.

Dosimetric comparison of interactive planned and dynamic dose calculated prostate seed brachytherapy

PURPOSE

To compare the dosimetrical results of an interactive planning procedure and a procedure based on dynamic dose calculation for permanent prostate brachytherapy.

MATERIALS AND METHODS

Between 6/2000 and 11/2005, 510 patients underwent (125)I implants for T1-T2 prostate cancer. Before 4/2003, 187 patients were treated using an interactive technique that included needle updating. After that period, 323 patients were treated with a more refined dynamic technique that included constant updating of the deposited seed position. The comparison is based on postimplant dose - volume parameters such as the V(100) and d(90) for the target, V(100)(r) for the rectum and d(10)(u) for the urethra. Furthermore, the target volume ratios (TVR identical with V(100)(body)/V(100)), and the homogeneity indices (HI identical with [V(100)-V(150)]/V(100)) were calculated as additional quality parameters.

RESULTS

The dose outside the target volume was significantly reduced, the V(100)(r) decreased from 1.4 cm(3) for the interactive technique to 0.6 cm(3) for the dynamic technique. Similarly the mean TVR reduced from 1.66 to 1.44. In addition, the mean V(100) increased from 92% for the interactive procedure to 95% for the dynamic procedure. More importantly, the percentage of patients with a V(100) < 80% reduced from 5% to 1%. A slight decline was observed with regard to the d(10)(u) (136% vs. 140%) and the HI (0.58 vs. 0.51).

CONCLUSION

The dynamic implant procedure resulted in improved implants. Almost ideal dose coverage was achieved, while minimizing the dose outside the prostate.

A study of a pretreatment method to predict the number of I-125 seeds required for prostate brachytherapy

PURPOSE

Prediction of the number of iodine seeds (I-125) required for prostate implantation is an important tool to reduce the number of unused seeds for brachytherapy. This study was designed to investigate the relationship between the number of seeds implanted vs. prostate volume. This can produce a tool to accurately estimate the number of seeds required for a given target volume. In addition, total cost of treatment, personal radiation risks during storage and handling, and errors in accounting for seeds can be reduced.

METHODS AND MATERIALS

Data from two groups of patients who had I-125 seed prostate implants (Oncura/Amersham RAPIDStrand model 6711 I-125) have been separately analyzed: (A) The relationship between prostate volume vs. number of seeds implanted was based on 401 patients treated between 1999 and 2002 who were implanted with seeds of air kerma strength (AKS) of 0.459 microGy h(-1) @ 1 m per seed. (B) The relationship between prostate volume vs. total seed AKS was analyzed. This was based on 628 patients treated between 1999 and 2002 who were implanted with a range of seed strengths from 0.381 to 0.521 U. Both patient groups were subdivided into integer prostate volume bins. For each bin, the mean and 95% confidence intervals (CI) for the implanted number of seeds or total AKS implanted were calculated. The upper 95% CI was used to investigate the relationship between the number of seeds implanted and total AKS implanted vs. prostate volume.

RESULTS

The new method of predicting the number of seeds shows valid and accurate results. The required number of seeds can be predicted, which helps to reduce the number of leftover seeds to 3% of the total number of seeds ordered.

CONCLUSION

The number of I-125 seeds or the total activity that is required to deliver the prescribed dose for the target volume can be predicted. This could reduce the overall treatment cost by accurate seed ordering before implantation.

PROSPECTIVE OUTCOMES ASSOCIATED WITH MIGRATION FROM PREOPERATIVE TO INTRAOPERATIVE PLANNED BRACHYTHERAPY: A SINGLE CENTER REPORT

PURPOSE

Brachytherapy (BT) for prostate cancer can be performed with either preoperative (PO) or intraoperative (IO) planned dosimetry. Potential benefits of intraoperative include fewer procedures for the patient, cost savings and improved accuracy of seed implantation leading to improved tumor control and urinary side effect profile. We report our experience with transition to IO planned BT after performing more than 600 PO planned implants since 1997.

MATERIALS AND METHODS

From September 2001 to February 2003, 46 consecutive patients with T1-3N0M0 adenocarcinoma of the prostate underwent BT. IO dosimetry was performed in 23 patients, while PO dosimetry was used in 23 immediately before changing to IO. American Urological Association (AUA) symptom index questionnaires were administered preoperatively and postoperatively. All patients underwent postoperative dosimetry by computerized tomography. Total, irritative and obstructive AUA scores were compared in the 2 groups using analysis of covariance. Models were adjusted for pretreatment variables of symptom scores, type of procedure and time since procedure.

RESULTS

Median followup was 47 and 45 days for PO and IO dosimetry, respectively. No differences were observed in seed, needle numbers or prostate size in the 2 groups. Average operative times were higher (47 vs 79 minutes, p <0.01) in the IO group but they decreased to nearly the same levels as PO implants in the first 23 cases so treated. Slopes of operative time over date of procedure differed significantly between methods (p <0.01). Comparing PO to IO dosimetry adjusted estimate of difference was -1.96 for total (95% CI -6.4, 2.5), -0.48 for obstructive (95% CI -3.3, 2.3) and -1.78 for irritative (95% CI -3.9, 0.31) AUA score. These differences were neither statistically nor clinically significant.

CONCLUSIONS

Our experience indicates that intraoperative planned BT is easily implemented in clinical practice as a result of a short learning curve. In addition, the approach is not associated with any changes in early postoperative voiding symptoms and, due to only marginally longer operative times, may have a cost advantage by eliminating the preplanning visit and ultrasound.

Prostate cancer brachytherapy: is real-time ultrasound-based dosimetry predictive of subsequent CT-based dose distribution calculation? a study of 450 patients by the Institut Curie/Hospital Cochin (Paris) Group

PURPOSE

Real-time ultrasound (US)-based dosimetry performed during (125)I loose seed implantation provides the radiation oncologist with an estimation of the dose distribution at seed insertion. However, for a number of reasons, this distribution may not reflect the real (reference) dosimetry as determined by subsequent CT, usually performed 1-2 months after implantation. The present study compared the two dosimetry data sets (US and CT) to evaluate how predictive extemporaneous US-based dosimetry can be of the real dose distribution.

METHODS AND MATERIALS

A total of 450 patients with prostate cancer were treated with loose (125)I seed implantation between June 1999 and October 2002 by the Institut Curie/Hospital Cochin (Paris) Group. The mean patient age was 65 years. Most patients (74%) had Stage T1c; the stage did not exceed T2b for the others. All patients had a prostate-specific antigen level of <15 ng/mL and was <10 ng/mL for 72%; 84% had a Gleason score of < or =6 and did not exceed 7 for the others; and 56% were treated with neoadjuvant hormonal therapy for a mean of 4.3 months. All patients were treated with loose seed implantation. Real-time US-based dosimetry was performed intraoperatively for all patients. CT-based dosimetry was performed 2 months after implantation, using the VariSeed software. The minimal dose to 90% of the outlined volume (D(90)) and percentage of volume receiving at least 100% of the prescribed dose (V(100)) were calculated with the two methods and compared for all patients.

RESULTS

On CT-based dosimetry, the D(90) was found to be > or =145 Gy (range, 115-240 Gy) in all patients except one. A large majority (86%) of patients showed a CT-based V(100) of >95%, and 48% had a V(100) of >98%. The mean CT-based D(90)/US-based D(90) ratio was 1.0 (range, 0.66-1.33). For 89% of the patients, the difference between the two values was <20% and for 62% was <10%. The mean CT-based V(100)/US-based V(100) ratio was 0.98 (range, 0-1.02), with 89% of patients showing a difference of <5%.

CONCLUSION

Our results indicate that the D(90) and V(100) values obtained intraoperatively with our real-time US-based dosimetry are in reasonable agreement with the subsequent values obtained with CT-based dosimetry performed 2 months after implantation. Recent innovations in our dose planning software allowed better control of the longitudinal seed position and could still improve the correlation between real-time US-based dosimetry and the subsequent CT-based dose distribution.