Validation of prospective whole-body bone marrow dosimetry by SPECT/CT multimodality imaging in ¹³¹I-anti-CD20 rituximab radioimmunotherapy of non-Hodgkin’s lymphoma

Abstract

Purpose

Radioimmunotherapy (RIT) for relapsed non-Hodgkin’s lymphoma is emerging as a promising treatment strategy. Myelosuppression is the dose-limiting toxicity and may be particularly problematic in patients heavily pretreated with chemotherapy. Reliable dosimetry is likely to minimise toxicity and improve treatment efficacy, and the aim of this study was to elucidate the complex problems of dosimetry of RIT by using an integrated SPECT/CT system.

Methods

As a part of a clinical trial of ¹³¹I-anti-CD20 rituximab RIT of non-Hodgkin’s lymphoma, we employed a patient-specific prospective dosimetry method utilising the whole-body effective half-life of antibody and the patient’s ideal weight to calculate the administered activity for RIT corresponding to a prescribed radiation absorbed dose of 0.75 Gy to the whole body. A novel technique of quantitation of bone marrow uptake with hybrid SPECT/CT imaging was developed to validate this methodology by using post-RIT extended imaging and data collection.

Results

A strong, statistically significant correlation (p=0.001) between whole-body effective half-life of antibody and effective marrow half-life was demonstrated. Furthermore, it was found that bone marrow activity concentration was proportional to administered activity per unit weight, height or body surface area (p<0.001).

Conclusion

The results of this study show the proposed whole-body dosimetry method to be valid and clinically applicable for safe, effective RIT.

Appendices

Appendix 1

Calibration of SPECT/CT scanner

For quantitation of ¹³¹I-rituximab uptake a SPECT/CT hybrid scanner (Millennium VG HawkEye, GE Medical Systems, WI, USA) was used [19]. To accurately quantify organ uptake of ¹³¹I using SPECT/CT imaging, a calibration factor must first be obtained to convert SPECT image count density (or count rate in the pixel or region) to the ¹³¹I tracer activity concentration for patient images, and thus in the organs, by extrapolation and summation.

For calibration, a standard reference phantom (PET/SPECT Performance Phantom 76-823, Nuclear Associates, NY, USA) with a “Linearity/Uniformity Insert” and a “Hot Lesion Insert” was filled with water and 403 MBq of ¹³¹I. The solution was mixed thoroughly to ensure uniform distribution of activity. The phantom was then scanned at 15 time points, each time point corresponding to a different activity concentration of ¹³¹I. The acquisition was performed using a standard ¹³¹I tumour imaging protocol comprising 90 projections (45 projections for each detector), at 50 s for every projection. The SPECT matrix was 128×128 and the pixel size, 3.45 mm. The CT component of acquisition used a full-circle rotation, 40 slices, with 140 kVp voltage and 2 mA current. Each scan was reconstructed using standard automatic iterative OSEM reconstruction on a nuclear medicine workstation (eNTegra v2.016, GE Medical Systems, WI, USA). All SPECT images were attenuation corrected by means of CT-based attenuation correction.

All SPECT and corresponding CT scans were reconstructed in 128 slices and then exported into Interfile image format. The number of counts was quantified in IDL (Research Systems, CO, USA) programming environment using several programs developed in-house. The calibration strategy comprised placing an ROI where counts or count density is not affected by partial volume effects, which may compromise application of calibration factors. The exclusion of partial volume effects in vitro was achieved by placing ROIs, where counting was performed, inside the phantom cylinder and ensuring that pixels near the walls of the phantom were not included by using an automated computer program written in IDL. The program scanned through the dataset on a slice-by-slice basis. First, the centre of mass in a slice was found and then a circle centred on the centre of mass was drawn in that slice to delineate the ROI. The circle had a diameter which was smaller than the physical diameter of the phantom by at least of one full-width at half-maximum, which was approximately 18 mm. The calibration VOI comprised all ROIs in all relevant slices. Such established count densities in VOIs were then plotted for 15 different activity concentrations. The calibration parameters were established by fitting a linear function onto measured data. The statistical errors were assessed using conventional parametric methods on goodness of fit and confidence intervals.

Calibration of linear function for computation of activity (kBq) in a pixel

$$ A_{{\text{pix}}} = aC + b $$

(4)

where A_pix is activity in a pixel in kBq, C is the number of counts in a pixel per second and a and b are calibration parameters. It should be noted that this is independent of pixel size. Calibration parameters as computed were a=8.0789±0.0579 kBq·s and b=0.0045±0.2109 kBq.

The SPECT/CT calibration is performed using a similar technique to that employed for SUV calibration of PET scanners. It is based on an ideal case of infinite source, that is, as if partial volume effect did not exist. Every implementation of such parameters thus may need to estimate a quantitation error. Errors will be dependent upon the volume and geometry of VOI and sources surrounding the VOI.

Appendix 2

Calculation of active marrow uptake

Uptake of ¹³¹I-rituximab in the bone marrow was estimated using marrow VOIs. It is reasonable to assume that the VOI drawn on CT consists of a mixture of cortical bone and bone marrow only. We can then represent the fraction of bone in the VOI as p and the fraction of marrow in the VOI as q, while M is the average intensity of the VOI on the CT image expressed in CT numbers. The mean CT number of bone tissue is 3,000 and mean CT numbers for red marrow and yellow marrow are 1,008 and 965, respectively [20, 21]. Thus the mean CT number of a mixture of yellow and red marrow will be dependent upon the ratio of yellow to red marrow. In practice, the bone marrow density and its CT number are dependent upon the fraction of yellow marrow, which will vary depending upon the location of bone marrow and the age of the patient. We used the values published in ICRP 23 [16] and ICRP 70 [18]. Active marrow volume fraction R is 0.7 (spinal), 0.48 (pelvic) and 0.25 (femoral), and the respective S numbers for spinal, pelvic and femoral marrow compartments are 995, 986 and 976.

Relationship between correction factors p and q

$$ 1 = p + q\quad M = 3000p + Sq $$

(5)

Computation of bone tissue correction factor q

Bone tissue fraction can then be expressed.

$$q=\frac{{M - 3000}}{{S - 3000}}$$

(6)

Formula for correcting activity concentration for bone tissue and yellow marrow volume

The activity concentration in active marrow (A_RM) can then be obtained from activity concentration in bone marrow VOI (A_{CT_VOI}) by correcting by factor q for supporting bone tissue volume and by factor R for yellow marrow volume.

$$ A_{{\text{RM}}} = \frac{{A_{{\text{CT\_VOI}}} }} {{qR}} $$

(7)