An alternative to the use of fixed activities is the administration of tracer levels of ¹³¹I to determine the characteristics of each patient, including the thyroid uptake and elimination half-times for radioactive iodine in the remaining thyroid tissue ^1. Such data may be used to establish more exactly the activity needed for ablation. To individualize the dosage in this manner, one needs to know with reasonable accuracy the mass of thyroid tissue to be treated, which is best evaluated with three dimensional imaging techniques. Knowledge of the distribution and retention of iodine in other tissues of the body is also desirable. The best image analysis techniques employ Single Photon Emission Computed Tomography (SPECT) or Positron Emission Tomography (PET), with anatomic images supplied by Computed Tomography (CT) or Magnetic Resonance Imaging (MRI) ^2, 3  devices.

In this work, we define a protocol for dose planning for thyroid ablation therapy in patients with differentiated thyroid cancer, using quantification of SPECT images. As such techniques may be difficult to implement in routine clinical use, we also show how a simple computer program can facilitate the application of such calculations with minimal data input by the user.

Calculation of Tissue Volume

Thyroid tissue volume was determined for each patient from the imate data; the volume of the remnant tissue used in the calculations was the largest value observed on the four images taken. After image processing, the calculated volumes were compared with the expected volumes, based on the difference between the standard volume of a normal thyroid (15-20 ml) and the volume of tissue removed during surgery, as determined during pathology examinations. As a further check on these calculations, images of each patient were transformed into GIF images and digitized with a program called SCMS (ref), developed to act as the interface between medical images (CT, SPECT or PET) and the Monte Carlo radiation transport code MCNP ^5. One of the modules of this code (SETUP) has a routine which can divide the volume of an organ into small regions. Then, the number of voxels in each region can be determined, and, knowing the size of each voxel (2.9 mm cube) the volume of the thyroid remnant tissue can be calculated analytically for each subject.

Determination of ¹³¹I Concentration

Using the reconstructed images, count densities were determined using Region-of-Interest (ROI) analyses in each slice of the image data, with the total count density determined as the sum of th densities in each slice, using calibration factors determined for the system (MBq/mL) (Lima et al. ^4). Knowing the activity concentration for each patient, we calculated the activity of ¹³¹I in the thryoid remants at the time of image acquisition. Thus the activity was calculated as the product of the activity concentration and the tissue volume.

Calculation of Optimal Therapeutic Activity

Activity values determined for each patient were plotted, and the uptake and elimination phases were fit to a two component exponential function using the Grace - 5.0 code. The fitted function was thus of the form:

where y(t) is the activity in the thyroid at any time, and a₀, a₁, and a₂ are fitted parameters. The cumulated activity is obtained by direct integration of equation (2), and the maximum activity in the thyroid is given as:

where f is the maximum fraction of administered activity (A₀) in the thyroid and C = [(a₂ / a₁)^{-a1/ (a2 / a1)} - (a₂ / a₁)^{-a2/ (a2 - a1)} ].

The standard dose equation for average dose in an organ is given as:

where m is the mass of the remnant thyroid tissue. Thus, the activity that should be administered to any subject is calculated as:

where Δ_β for ¹³¹I is 3.07 x 10^-14 Gy.kg/Bq.s.

Computer Program for Dose Planning

To facilitate the implementation of the above considerations in routine nuclear medicine practice, we created a compute program, called PlanDose, which provides the optimum therapeutic activity of ¹³¹I to be administered to a patient, given the input of appropriate data. The program was written in C, which allows for flexibility and portability on many platforms when compiled. Users must enter data from the biokinetic study for a subject, and the program fits the data to the function shown in Eq. (2), using a Figure of Merit analysis (χ_i²)¹:

where, y_i is a measurement-dependent variable and f(x_i) is the value of the variable which depends on the point x in the model. The variables a₀, a₁ and a₂ are obtained through fitting and minimization of this function. To treat a nonlinear function, the minimization is done through an iterative (Levenberg-Marquardt) model. The minimization criteria are that (χ²_i-χ²_i+1)² < 0,001 or N = 40 where N is the number of iterations. These values were selected to limit the time needed to fit the data using Grace 5.0. Then, once the fitted coefficients of the time-activity curve are obtained, the maximum thyroid uptake and therapeutic activity needed for the subject are calculated according to equations (3) and (5).

Table 1 shows a comparison of thyroid remnants volumes between the measured surgical specimens and the values calculated using SPECT images with ¹³¹I. Table 2 gives the values calculated using the SCMS program.

Table 1. Thyroid remnant volumes estimated by SPECT and surgical
specimens in patients undergoing thyroidectomy.

^*Exact dimensions of the pathological sample were not known, as only a lobectomy was performed. Value deduced by consideration of the remnant in the other lobe.Volver>>

Table 2. Comparison of thyroid remnant volume between the value estimated by SPECT and the SCMS method.

Volver>>

Activity Concentration

Table 3 gives the concentrations of activity observed at various times post administration in the patient population studied. The concentrations varied between 0.02 MBq/mL and 7.23 MBq/mL. Using these measurements and knowing the remnant volume in each case, the iodine activity in the remnant tissue may be calculated at each time. Figure 1 shows the results for one subject after administration of 148 MBq (4 mCi) of ¹³¹I. These data are typical of iodine retention after thyroidectomy.

Table 3. ¹³¹I activity concentration in thyroid remnants of patients with CDT.

Figure 1. Kinetic model for 131I in thyroid remnant tissue for one subject, after administration of 148 MBq (4 mCi).  Volver>>

Using the retention curve for each subject, we calculated the maximum thyroid activity values, which varied between 0.06 and 20.7 MBq (1.6 - 560 ìCi), corresponding to maximum percent uptake values between 0.04% and 13.8%, with a mean value of 3.8%. These values agree with the findings of others in similar studies ^{8, 9, 10}.

Therapeutic Activity

The best therapeutic activity for each subject was calculated using the data above, and assuming a desired therapeutic absorbed dose of 300 Gy, as suggested by various authors ^{11, 12, 13}. These results are shown in Table 4.

Table 4. Therapeutic activities suggested for the subjects in this study, assuming a desired dose of 300 Gy to the thyroid remnant tissue. Volver

The PlanDose Program

The program requires as input the results of the study of the individual subject’s biokinetics, i.e. the activity measured at various times after administration of the tracer quantity of ¹³¹I. The user must also provide the mass of the thyroid remnant tissue and the amount of ¹³¹I administered in the tracer study. The user may also enter some limited data for patient identification, as is shown in Figure 2. The program uses units for input and output as are commonly used in most nuclear medicine centers currently. The program fits the entered data to the functional form of Eq. (2), and calculates the maximum thyroid uptake and therapeutic activity recommended for the subject. The program also shows the figure of merit observed for the data (Figure 3).

Figure 2. Form for the PlanDose Program, showing the input data for one thyroidectomy patient

Figure 3. Form for the PlanDose Program, showing the results of the curve fit to the entered data and the calculation of suggested therapeutic activity

 Table 2 shows that the percent difference between the volumes calculated by SPECT volumetrics and the technique using the SCMS program were on average 12.8 ± 1.8%. Lima et al. ⁴ found a similar accuracy (10.2%) for sources of irregular geometry. The methodology shown here provides a reasonable accuracy (around 10%), given the limits of resolution of the imaging system and other uncertainties in routine dose calculations.

Figure1shows that the maximum thyroid remnant activity occurs at about 8 h post administration of ¹³¹I. This time varied between 8 and 10 h, similar to values observed by Johansson et al (private communication)*. Use of the more realistic two component retention curve (compared to the classical, one compartment model), results in a better estimation of the optimum ¹³¹I activity to be administered. Use of more simplistic methods has been shown to possibly overestimate the activity to be administered by up to a factor of three ^7.

Noting that all of the subjects received therapeutic activities of 3.7 GBq (100 mCi), Table 4 shows that the majority (7 of 9) could have received less activity, thus receiving lower doses to other organs, especially marrow and gonads. The reduction in marrow dose will be due to a lower level of ¹³¹I in the circulating blood, which also contributes to dose in other tissues. The dose to the gonads will be lower, primarily due to lower amounts of activity being eliminated through the urine. Such dose reductions justify the use of methods which optimize the therapy design for individual subjects.

Only two subjects (G and H) would not have had their activity levels reduced. The therapeutic activity levels suggested by the method were not practical; this is most likely due to these subjects showing very low thyroid remnant uptake values (< 1,0%). These levels of concentration were not treated in the study of correlations between count and activity concentrations by Lima et al. According to Mortelmans et al. ¹⁵  (1986) activity levels below a certain threshold cannot be accurately determined due to limitations in image contrast. In such cases, the concentration should be determined using a subtraction method using the maximum counts in the supplied images. It is likely that this method should be used for concentrations less than 0.6 MBq/mL (16 ìCi/mL). In these cases, patients should not receive doses higher than the routing fixed level of 3.7 GBq (100 mCi).

The data here show that for three of the patients, no hospitalization is necessary, as they could have received as little as 1.1 GBq (30 mCi). This would reduce costs and provide the subjects with an improved quality of life, as they can return to their homes immediately after administration of the therapy. Becker et al¹ were able to treat about 40% of their subjects as ambulatory, using reduced activity levels (740-4810 MBq (20-130 mCi)), with therapeutic effectiveness similar to that using fixed activity values. Their methods did not treat the uptake period, but assumed an “instantaneous” uptake value at 24 hours post administration, which led to an overestimate of absorbed dose in the thyroid remnant tissue. Thomas et al¹⁶ found a dose overestimate of only 10%-15%, which they considered acceptable.

The nuclear medicine clinic as well will benefit from lower administered activity values, as they can use less radioactive material over time, which will reduce costs as well as radiation dose to the clinic staff. The PlanDose program facilitates the application of this technique for the clinic staff. Expertise in radiation dosimetry is not required, only the accurate entry of the subject’s biokinetic data. The target radiation dose value of 300 Gy is also an adjustable parameter in the program, so that the physician can vary this as desired. As with any computer program, however, care must be taken in the evaluation of the output, with both a physicist and the attending physician carefully studying the results and making the final recommendation on the desired level of therapeutic activity to be administered.

In our experience, the Grace 5.0 converged rapidly on an optimum fit in all cases. In one case, however, the fit was interrupted as it reached the maximum number of iterations. Computer fitting of measured data must also be reviewed critically by the professional staff in all cases to ensure that reasonable values are used in the final analysis.

Although this work considered only thyroid ablation, the principles could be extended to the treatment of hyperthyroidism. As shown by Traino et al. ^17, patient-specific metabolic data and organ mass data obtained via ultrasound measurements can facilitate the calculation of subject-specific therapeutic activity levels. The thyroid masses can also be measured using SPECT methods, although these methods may be less reliable, considering the resolution limits of current cameras and the appearance of the “functional volume” being larger than the actual tissue volume.

___________________________________