I read with interest the article by Levesque et al1 in the June/July issue of the AJNR reporting on recanalization of animal model arteries after embolization with 32P-oligodeoxynucleotide-coated Gugliemi coils. In their report, the authors claim to have developed a method to bind tightly a 32P 15 mer oligonucleotide to Guglielmi platinum coils that has some advantages compared with ion-implanted coils proposed by an independent group of investigators.2 The technique they describe consists of simply “dipping” the platinum coils in a 32P-oligonucleotide solution with the coil surface adsorption coefficient varying with the solution’s temperature. Variability in the total activity attained by this technique as a function of the solution’s temperature and the coil length are illustrated in Figures 1and 2 of their report and indicate a SEM of about ±10%, which appears reasonable. Thus, it is tempting to conclude that their new technique for the production of 32P-coated coils offers ease of production, good accuracy, and reproducibility. Nevertheless, using the SEM instead of the SD in presenting the data can be misleading. This is because the SEM gives an idea of the accuracy of the mean value of a population, whereas the SD gives you an idea of the variability of single observations. The 2 are related by SEM = SD/(square root of sample size). Moreover, a 95% confidence interval in the measured quantities is represented by 2 SD. Using this latter value yields a variability of single observations of greater than ±45% (95% confidence interval) assuming the mean is averaged ≥5 experiments. This more meaningful quantity contrasts significantly with the approximately 10% error bar reported in Fig 2 of Levesque et al’s report.1
Human experiments by using 32P-ion-implanted coils have been reported recently. In these experiments, the coils were ion-implanted, a technique that physically binds the 32P atoms to the metallic surface with negligible leaching. Ion implantation yields coils with activities that can be measured accurately within ±5% (SD) by using standard counting techniques. Also, because there is no leaching of the 32P, the dose can be predicted accurately (usually within ±10% SD) because the radioactive decay from the coil surface is due entirely to the physical decay with a half-life of 14.3 days for 32P. This level of accuracy is within the typical standards usually found in intravascular brachytherapy (IVB), and is critical for the safety and potential success of the 32P coil treatment in humans.2
In contrast, a 32P-“dipped” coil is comparable to a local drug delivery device where the coated drug is slowly eluted from the device in a less predictable manner, resulting in large uncertainties in the coil activity, distribution, and residence time of the 32P at the target site. This results in large uncertainties in predicting the radiation dose that is actually delivered to the tissues. The reasons for this have been expressed in detail in a recent publication about the dosimetry of 32P- oligonucleotide-“dipped” stents for the treatment of restenosis.3 In brief, a conventional IVB source (eg, 32P-ion-implanted stents or coils without any leaching) is characterized by physical factors that can be measured accurately (eg, activity, geometry), which yield accurate dose calculations. In contrast, for a drug-eluting device (stent or coil), the amount of radioactivity deposited in the tissues is strongly dependent on biologic factors (drug uptake, washout rate, residence time, diffusion) that can fluctuate significantly from patient to patient. This makes it virtually impossible to predict accurately the dose that is delivered to the tissue during the experiment (not even within ±50%), falling short of the standards of quality of conventional brachytherapy. Drug leaching from dipped coils will also result in a small but unnecessary dose to healthy organs. On the basis of these arguments, we conclude that 32P-oligonucleotide-dipped coils are not a valid alternative to ion-implanted coils. Their usage in humans is risky, because it is unlikely that the prescribed radiation dose can be delivered accurately and effectively because of the low predictability and reproducibility of the drug-elution parameters.
Reply:
We are grateful to Dr. Janicki for showing in interest in our recent article in the AJNR.1 Ion implantation does provide a means to better fix the isotope onto the coils, though some in vitro and in vivo leaching is still possible.2 We doubt that leaching of a fraction of activities of 32P prescribed to prevent recanalization, with subsequent biodistribution, represents a definite health hazard, but a policy of minimizing such unnecessary exposure is certainly prudent. Perhaps more important, one can rely on ion-implanted coils to deliver and keep activities at the target site with more accuracy and better assure the efficacy of the strategy. In light of the steep nature of the dose/distance curve obtained with beta radiation, the uncertainties regarding the anatomy of the target tissues and the impossibility of determining a priori the exact position of the coils before their in vivo deployment, any in situ beta radiation strategy will always involve difficulties in dose calculations. Nevertheless, approximations are possible—no matter their exact disposition, most coils will be confined to the aneurismal sac—and perhaps sufficient to prescribe activities according to a “therapeutic window.”3, 4; Whether the risks involved with such approximations are worth taking depends on the expected benefits of the strategy and the comparative efficacy and risks of clinical alternatives.
The article describes a method to circumvent the problem of the half-life of 32P (2 weeks) entailing conceptual difficulties in the management of coil inventories. The challenge is to deliver active coils promptly in centers throughout the world. Other methods to do so while conserving the advantages of ion implantation may exist, but they involve other difficulties.4 No matter how effective a new treatment may be in the laboratory, it cannot have any clinical impact if it does not reach the patient for whom it was designed. It is an unfortunate but uncontrollable fact that new devices will not be put into clinical use unless they entail profits to an organization. Nevertheless, we have failed thus far to convince the industry that ion implantation of endovascular coils could be a profitable enterprise. There are many drawbacks to the method described, but it did provide coils that were effective in preventing recanalization in experimental arterial occlusion models, a feat that remains unchallenged by coils available on the market.
- Copyright © American Society of Neuroradiology