Research ArticleClinical Investigation

SPECT/CT Image-Derived Absorbed Dose to Red Marrow Correlates with Hematologic Toxicity in Patients Treated with [177Lu]Lu-DOTATATE

Johan Blakkisrud, Avery B. Peterson, Scott J. Wildermann, Griffen Kingkiner, Ka Kit Wong, Chang Wang, Kirk A. Frey, Caroline Stokke and Yuni K. Dewaraja
Journal of Nuclear Medicine May 2024, 65 (5) 753-760; DOI: https://doi.org/10.2967/jnumed.123.266843

Abstract

Hematologic toxicity, although often transient, is the most common limiting adverse effect during somatostatin peptide receptor radionuclide therapy. This study investigated the association between Monte Carlo–derived absorbed dose to the red marrow (RM) and hematologic toxicity in patients being treated for their neuroendocrine tumors. Methods: Twenty patients each receiving 4 treatment cycles of [177Lu]Lu-DOTATATE were included. Multiple-time-point 177Lu SPECT/CT imaging–based RM dosimetry was performed using an artificial intelligence–driven workflow to segment vertebral spongiosa within the field of view (FOV). This workflow was coupled with an in-house macroscale/microscale Monte Carlo code that incorporates a spongiosa microstructure model. Absorbed dose estimates to RM in lumbar and thoracic vertebrae within the FOV, considered as representations of the whole-body RM absorbed dose, were correlated with hematologic toxicity markers at about 8 wk after each cycle and at 3- and 6-mo follow-up after completion of all cycles. Results: The median of absorbed dose to RM in lumbar and thoracic vertebrae within the FOV (Dmedian,vertebrae) ranged from 0.019 to 0.11 Gy/GBq. The median of cumulative absorbed dose across all 4 cycles was 1.3 Gy (range, 0.6–2.5 Gy). Hematologic toxicity was generally mild, with no grade 2 or higher toxicity for platelets, neutrophils, or hemoglobin. However, there was a decline in blood counts over time, with a fractional value relative to baseline at 6 mo of 74%, 97%, 57%, and 97%, for platelets, neutrophils, lymphocytes, and hemoglobin, respectively. Statistically significant correlations were found between a subset of hematologic toxicity markers and RM absorbed doses, both during treatment and at 3- and 6-mo follow-up. This included a correlation between the platelet count relative to baseline at 6-mo follow up: Dmedian,vertebrae (r = −0.64, P = 0.015), Dmedian,lumbar (r = −0.72, P = 0.0038), Dmedian,thoracic (r = −0.58, P = 0.029), and Daverage,vertebrae (r = −0.66, P = 0.010), where Dmedian,lumbar and Dmedian,thoracic are median absorbed dose to the RM in the lumbar and thoracic vertebrae, respectively, within the FOV and Daverage,vertebrae is the mass-weighted average absorbed dose of all vertebrae. Conclusion: This study found a significant correlation between image-derived absorbed dose to the RM and hematologic toxicity, including a relative reduction of platelets at 6-mo follow up. These findings indicate that absorbed dose to the RM can potentially be used to understand and manage hematologic toxicity in peptide receptor radionuclide therapy.

Peptide receptor radionuclide therapy (PRRT) with somatostatin analogs is an established treatment option for somatostatin receptor–expressing neuroendocrine tumors (1) The combination of the analog octreotate and the radionuclide 177Lu, known as [177Lu]Lu-DOTA0-Tyr3-octreotate ([177Lu]Lu-DOTATATE), received approval from the Food and Drug Administration in 2018.

[177Lu]Lu-DOTATATE is generally well tolerated, with the kidneys and red marrow (RM) considered the dose-limiting organs. Transient occurrence of grade 3 or 4 hematologic toxicity has been observed in approximately 5%–10% of patients in previous studies, potentially leading to treatment posology, postponing, or termination (1–4).

The hematologic toxicity could be attributed to irradiation of the RM, as mechanistically certain groups of marrow progenitor cells express somatostatin receptors (5). An alternative explanation is binding of transferring and free 177Lu (6). Previous work with RM dosimetry has included both direct, image-based, methods and blood-based methods in which the blood is used as a surrogate. If there is specific uptake, a blood-based method will most likely underestimate the absorbed dose to RM regardless of the underlying mechanism. Early studies investigating diagnostic scans with [86Y]Y-DOTATOC before therapy with [90Y]Y-DOTATOC unveiled variable uptake in the RM among patients (7). The patient-specific variability motivates the use of image-based dosimetry for treatment personalization.

Image-based methods for absorbed dose calculation to the RM pose 2 main challenges. Since RM is a distributed organ, ideally the absorbed dose in every part of the skeleton assumed to contain RM should be calculated; however, this process is time-consuming and often unfeasible because of lack of whole-body posttherapy SPECT/CT images. Alternatives include using a subset of the skeletal sites; for example, use of the lumbar vertebrae as a representation of the whole marrow has previously been suggested (8,9). Recent developments in artificial intelligence–driven methods (10) have made segmentation of larger parts of the skeleton more feasible as an integrated part in the dosimetry workflow.

Furthermore, the microdistribution of the RM within skeletal sites presents an additional challenge in the dosimetry calculation because of the complex composition of the spongiosa, which comprises trabecular bone, yellow marrow, and RM. Efforts have been made to develop dosimetry models for spongiosa dosimetry that aim to simulate the microstructure of the marrow spongiosa, including RM as a source and target organ (11,12). Coupling of these spongiosa models with functional imaging enables calculation of absorbed doses to the RM.

Accurate calculation of absorbed doses becomes crucial in understanding the relationship between radiation exposure and myelosuppression. Although previous studies on PRRT have identified potential associations between absorbed dose and toxicity, the observed correlations have been relatively weak and limited to toxicity at nadir (13). The aim of the current work was to investigate the absorbed dose to the RM during multiple treatment cycles of [177Lu]Lu-DOTATATE PRRT using a state-of-the-art SPECT/CT dosimetry workflow including an in-house Monte Carlo (MC) dosimetry code that accounts for radiation interaction in the spongiosa microstructure. Our primary focus was to evaluate subacute and chronic myelosuppression and the association with RM absorbed dose.

MATERIALS AND METHODS

Twenty patients with histologically confirmed neuroendocrine carcinoma who were treated with [177Lu]Lu-DOTATATE at the University of Michigan Hospital between July 2020 and January 2022 were included in this study. Internal review board approval and written informed consent was obtained for the SPECT/CT imaging. The treatment protocol consisted of administering a nominal activity of 7.4 GBq per treatment fraction, with a total of 4 treatment fractions spaced approximately 8 wk apart. The mean of the administered activity per treatment fraction was 7.3 GBq (range, 6.8–7.6 GBq). Table 1 summarizes the patient characteristics.

View this table:
TABLE 1.

Patient Characteristics

An overview of the dosimetry calculation is given in Figure 1.

FIGURE 1.
FIGURE 1.

Overview of RM absorbed dose calculation from SPECT/CT images, via segmentation, registration, and time–activity fitting/integration and microscale/macroscale MC simulation. Arrows in example patient SPECT/CT images show vertebrae with focal uptake due to bone lesion that was not included in analysis. Created with BioRender.com.

Quantitative SPECT/CT Imaging

177Lu SPECT/CT images were acquired using an Intevo Bold scanner (Siemens Healthineers) with a single field of view (FOV) set to cover the abdominal area. Images were acquired after each of 4 treatment cycles and for most patients at 4 time points: on day 0 (day of administration), day 1, days 4–5, and days 5–8 after administration. The imaging and reconstruction protocol has been previously described (14). In short, images were reconstructed with 48 iterations (1 subset) of ordered-subset conjugate gradient with triple-energy-window scatter correction, CT-based attenuation correction, resolution recovery, and no postfiltering. Non–contrast-enhanced free-breathing CT was performed at 120 kVp and 80 mAs as a reference scan, usually on day 1, whereas 15 mAs were used for the remaining time points.

Spongiosa Segmentation

The CT image obtained on day 1, considered the reference time point, was used to segment the skeletal sites for dosimetry calculations. Initially, the TotalSegmentator software tool, based on the nnU-NET neural network, was used to perform an automated segmentation of the vertebrae within the FOV (10,15). Additional automated postprocessing steps were then undertaken to eliminate the cortical bone surrounding the spongiosa: first, a subthreshold of 400 Hounsfield units was applied, and subsequently, a series of morphologic operations was conducted to remove islands and salt-and-pepper noise. The resulting spongiosa segmentations underwent manual verification and adjustment if necessary to ensure their accuracy.

The SPECT images were registered to the reference time point using the SPECT–SPECT contour intensity-based registration algorithm (14) from MIM Software, with the spongiosa volume serving as the target structure.

Time-Integrated Activity Map

Time–activity curves were generated by fitting the sum of activity in each aligned vertebra using a piecewise approach (Supplemental Fig. 1; supplemental materials are available at http://jnm.snmjournals.org (16,17)) to account for high noise associated with a low activity concentration. A monoexponential function was fitted to the initial time points, if the total activity of the structure was above 0.3 MBq, followed by trapezoidal integration for activity below this cutoff. This threshold was based on a prior noise study with a 177Lu phantom (18). Beyond the last time point, a monoexponential fit function with physical half-life was assumed. The resulting time-integrated activity was then placed within each vertebra with an interior distribution based on the activity distribution from the SPECT scan at the reference time point. The mean time–activity data for the remainder of body (whole body except vertebrae) was fit with a biexponential curve by the MIM software, and the time-integrated activity was distributed such that the activity distribution at the reference time point was retained.

For cycles with fewer than 4 available time points, a calculation incorporating kinetic information from prior or other cycles was used. This approach was used only when a scan was available on day zero, as later time points were deemed unreliable because of a low activity concentration. The ratio of the activity on the available scan on day zero and the time-matched point on the previously computed time–activity curve for either the preceding or—in the case of treatment cycle 1, subsequent—treatment fraction was calculated. This ratio was then applied to the previously calculated time-integrated activity map by multiplying the voxels within each vertebral structure by the corresponding ratio. The ratio for the volume outside the structures was determined by comparing the sum of activity of the whole image outside the vertebral volumes.

MC Radiation Transport with Microlevel Calculations in Spongiosa

For defining the patient geometry for use in the radiation transport simulation, a density map was made from the reference CT image using a system-specific CT-density calibration curve. The time–activity map, spongiosa volumes of interest, and density map were used as inputs to the in-house MC code Dose Planning Method (DPM) (19). For each MC run, a total of 100 million histories was used to keep statistical uncertainty to less than 2% in RM regions.

The voxel-level (macroscale) DPM MC code was adapted to compute doses of spongiosa components by incorporating energy absorption fractions that had been precomputed on the subvoxel (microscale, 50-μm voxel size) level for electrons originating from or entering regions defined to be spongiosa structures. This approach has previously been implemented for 131I (12), and DPM was recently extended to 177Lu with a modified radionuclide source (14).

The microscale model has the option to distribute cumulative activity in spongiosa voxels among 3 components: the RM, yellow marrow, and trabecular bone. In this work, it was assumed that [177Lu]Lu-DOTATATE was attached to the RM cells, as this has been indicated by prior studies (20). The activity was distributed uniformly in the RM subvoxels within each SPECT voxel. Electrons in the spongiosa are assumed to deposit energy locally in the 3 constituent components in proportion to precalculated electron energy absorption fractions. Tabulated values of energy absorption fractions were calculated by the general-purpose MC code Electron-Gamma Shower 5 for a range of bone volume fractions (BVFs) and cellularity factors (CFs [fraction of marrow occupied by RM]) (12) as a function of the site of electron emission. With the 3-component model of the spongiosa, the voxel-specific BVF can be calculated for each voxel on the basis of an assumed CF, the CT-based voxel density (Embedded Image), the densities of red (Embedded Image g/mL) and yellow (Embedded Image g/mL) marrow, the density of bone (Embedded Image g/mL), and the following equations:Embedded ImageEmbedded Image

The total mass of RM in each vertebra was also calculated by summing the marrow content of each voxel i with volume Embedded Image:Embedded Image

An International Commission on Radiological Protection publication 70 reference CF of 70% was used for the absorbed dose calculation, and the tissue densities were taken from International Commission on Radiological Protection publication 110 (21,22).

RM Absorbed Doses

Since the entire spongiosa was not captured by SPECT/CT imaging, the absorbed dose to vertebrae in the FOV was used to represent the absorbed dose to the whole marrow. Vertebrae with focal uptake, assumed to be due to bone lesions, as well as vertebrae only partially in the FOV were excluded. Three dose metrics, Dmedian,lumbar, Dmedian,thoracic, and Dmedian,vertebrae, were defined as the median absorbed dose to the RM in the thoracic, lumbar, and all vertebrae within the FOV, respectively. Additionally, the mass-weighted average absorbed dose of all (N) vertebrae (Daverage,vertebrae) were calculated asEmbedded Imagewhere Dvertebrae,n and MRM,vertebra,n are the RM absorbed dose and RM mass of an individual vertebra, respectively. Each of the 4 dose metrics was used separately in the dose–toxicity analyses.

Dose–Toxicity Correlations

Laboratory blood count data were available from patient medical records for various time points, including baseline (before the first treatment cycle), before each subsequent cycle (∼8 wk after prior cycle), and at 3- and 6-mo follow-up after the last treatment cycle. The analyzed blood values included absolute platelet count, absolute neutrophil count, absolute lymphocyte count, and hemoglobin. Evaluation of the toxicity markers was based on both the absolute values and their relative value (ratio) compared with baseline. The absolute toxicity values were graded according to the Common Terminology Criteria for Adverse Events, version 5.0 (23). For dose–toxicity correlations assessed between treatment cycles, the correlation analysis included the sum of absorbed dose up to the respective toxicity measurement. For dose–toxicity correlations assessed at 3- and 6-mo follow-up, the correlation analysis considered only the cumulative absorbed dose to the RM from all 4 treatment cycles.

Multivariable Analysis

In addition to exploring the absorbed dose to the RM, this study aimed to investigate the potential influence of baseline patient characteristics on the dose–toxicity relationship. Bivariable models were created in which the RM absorbed dose and the baseline characteristics were included both as main effect terms and as interacting terms. Bonferroni adjustment was used. A list of explored patient characteristics is given in Supplemental Table 1.

Statistics

Data were analyzed for normality with the D’Agostino and Pearson omnibus test. Because there was a mixture of normal and nonnormal distributed data, nonparametric tests were used throughout. Correlation between absorbed dose and hematologic toxicity was explored with the Spearman rank test. The Wilcoxon signed-rank test was used to pairwise-compare the absorbed dose across cycles in the same patient, as well as the absorbed doses to different vertebrae regions in the same patient. For all statistical analysis, the significance level was a P value of 0.05. All statistical calculations were done with the Python statistics library scipy stats (scipy, version 1.9) and R, version 4.1.2.

RESULTS

Absorbed doses to RM were calculated for 20 patients across all 4 treatment cycles, resulting in 80 cycles in total. For 10 of 80 cycles, a prior information dosimetry approach was used because of missing SPECT/CT time points. Focal uptake in the vertebrae led to the exclusion of 18 vertebrae among 7 patients from the analysis. Two of the patients included in the present work had previously received PRRT. A recent analysis investigating safety profiles of PRRT in a retreatment setting has indicated similarity to initial PRRT measures (24). Consequently, these retreated patients were included in the cohort. The mean Dmedian,vertebrae per treatment cycle was 0.34 Gy (0.14–0.78 Gy), and the median cumulative value across all 4 cycles was 1.3 Gy (0.6–2.5 Gy). Additionally, the mean absorbed dose to the RM per administered activity for all treatment cycles expressed by Dmedian,vertebrae was 0.049 Gy/GBq (0.019–0.11 Gy/GBq). Figure 2 displays RM absorbed dose values for each patient and each cycle. No systematic differences were observed from cycle to cycle within the same patient (Wilcoxon signed-rank, P > 0.43 for all comparisons). There was no adjustment of treatment interval due to hematologic toxicity. Unrelated to hematologic toxicity, patient 23 had an interval of 8 mo between treatment cycles 2 and 3.

FIGURE 2.
FIGURE 2.

Vertebral absorbed doses with patient median and 95% CI. Pnum = patient number.

The RM absorbed doses to lumbar and thoracic vertebrae exhibited strong correlations (Pearson r = 0.86–0.98), and the differences between them are depicted in the Bland–Altman plot shown in Figure 3. The lumbar vertebrae displayed a bias of 0.018 Gy when compared with the median of all vertebrae, whereas the thoracic vertebrae exhibited a bias of −0.017 Gy. A statistically significant difference was observed between the averaged dose across all skeletal sites and the lumbar dose (Wilcoxon signed-rank, P = 0.01) but not between any of the other pairs.

FIGURE 3.
FIGURE 3.

Bland–Altman plot illustrating differences between dose metrics Dmedian,lumbar (A), Dmedian,thoracic (B), and Daverage,vertebrae (C) compared with Dmedian,vertebrae. Each data point represents individual treatment cycle.

Figure 4 presents the intercycle and follow-up toxicity data. For some patients, laboratory blood values were unavailable for the 3- and 6-mo follow-up (Table 1). No thrombocytopenia or neutropenia of grade 2 or worse was observed among the included patients. Grade 3 absolute lymphocyte count toxicity occurred in 4 patients at some point during treatment or follow-up. No grade 4 toxicity was recorded for any of the hematologic toxicity markers at the sampling time points.

FIGURE 4.
FIGURE 4.

Box plots describing platelet count (A), absolute lymphocyte count (B), hemoglobin (C), and absolute neutrophil count (D) sampled before each treatment cycle and at 3- and 6-mo follow-up. Overlaid swarm plots are with individual datapoints. Dashed, dashed–dotted, and solid lines indicate levels for Common Terminology Criteria for Adverse Events of grades 2, 3 and 4, respectively. ALC = absolute lymphocyte count; ANC = absolute neutrophil count; HB = hemoglobin; PLT = platelet count.

During treatment, 14 combinations of RM absorbed dose metrics and hematologic toxicity markers were identified as significant for dose–toxicity correlations (Table 2). Notably, the platelet count before treatment cycle 2 (8 wk after cycle 1) relative to baseline was negatively correlated with every RM absorbed dose metric for cycle 1: Dmedian,thoracic (r = −0.47, P = 0.038), Dmedian,vertebrae (r = −0.51, P = 0.034), Dmedian,lumbar (r = −0.48, P = 0.034), and Daverage,vertebrae (r = −0.50, P = 0.025). The platelet count before cycle 4 relative to baseline was also negatively correlated with the RM absorbed dose metrics, except for Dmedian,vertebrae (r = −0.41, P = 0.07).

View this table:
TABLE 2.

Significant Correlations Between Absorbed Dose Metrics and Hematologic Toxicity

For follow-up, 12 pairs of combinations of RM absorbed dose and toxicity parameters were found to be significant for dose–toxicity correlations (Table 2). In general, the correlations between the relative platelet count compared with baseline became stronger at the 6-mo follow-up than between treatments. This held true for all absorbed dose metrics: Dmedian,vertebrae (r = −0.64, P = 0.015), Dmedian,lumbar (r = −0.72, P = 0.0038), Dmedian,thoracic (r = −0.58, P = 0.029), and Daverage,vertebrae (r = −0.66, P = 0.010). In addition, the absolute value of absolute lymphocyte count correlated with Dmedian,vertebrae (r = −0.55, P = 0.022), Dmedian,lumbar (r = −0.53, P = 0.029), and Daverage,vertebrae (r = −0.55, P = 0.023). The absolute value of platelets at 3 mo also significantly correlated with all absorbed dose metrics. Scatterplots of the Daverage,vertebrae and platelet count and absolute neutrophil count reduction relative to baseline before the second and last treatment cycles as well as 6-mo follow-up are shown in Figure 5.

FIGURE 5.
FIGURE 5.

Relationship between averaged absorbed dose and relative number of platelets (A–C) and neutrophils (D–F) compared with baseline. (A and D) Absorbed dose in treatment cycle 1 correlated with relative blood counts after first cycle. (B and E) Sum of first 3 absorbed dose to RM and relative blood count value before last treatment cycle. (C and F) Cumulative absorbed dose from all 4 cycles and relative blood count value 6 mo after last treatment. ANC = absolute neutrophil count; PLT = platelet count.

In the multivariable analysis, previous chemotherapy before PRRT was a significant interacting parameter together with RM absorbed dose (P = 0.01) but not under Bonferroni correction (threshold, 0.0025); the RM of patients who had received prior treatment was more radiosensitive (Supplemental Table 1).

DISCUSSION

We assessed the RM absorbed dose in patients treated with [177Lu]Lu-DOTATATE and found a significant correlation with hematologic laboratory levels, particularity platelets. This correlation was observed during the treatment course and persisted at 3- and 6-mo follow-up.

The overall incidence of hematologic toxicity was low, most probably because the nadir of the toxicity typically occurs between cycles, approximately 4–6 wk after activity administrations, as indicated in previous studies (4), whereas the blood count measurements in our study occurred at about 8 wk. In a recent study, which used quantitative imaging, RM S values, and an assumption of specific uptake, the reported RM absorbed doses of 0.056 ± 0.023 Gy/GBq for vertebrae (T9–L5) without metastases closely aligns with our results (0.049 ± 0.019 Gy/GBq) (20). Hagmarker et al. investigated the dose–toxicity relationship using absorbed dose to the entire spongiosa as a representation of the RM (13). Although direct comparison is not possible because of their definition of toxicity (at nadir), they also reported an association between absorbed dose and myelosuppression. Despite the differences in focus, our combined findings demonstrate that the dose–toxicity relationship extends beyond the acute nadir period, encompassing longer-term toxicity as well.

Previous studies have demonstrated a correlation between absorbed dose to the spleen and hematologic toxicity, potentially due to the spleen’s role as a reservoir for blood elements (25). Additionally, correlations between hematologic toxicity and renal function (4,26), patient age (4), and tumor burden (4,26) have been observed. In the current work, we identified a significant albeit weak interaction between previous systemic treatment and absorbed dose. However, this did not pass the Bonferroni correction, and none of the other patient characteristics included in our analysis improved the correlations within our present cohort (Supplemental Table 1). The limited size and heterogeneous composition of our patient group prevent us from drawing conclusive results on these additional parameters.

Conventionally, a 2-Gy limit to RM exposure has been used as an adjustment criterion in personalized treatment studies with PRRT (27,28). Caution is advised in strictly adhering to this limit, as the original work (29) used blood as a surrogate organ for RM and treatment of differentiated thyroid cancer with 131I-NaI, involving significant differences from DOTATATE PRRT in dose rate and fractionation. Our findings indicate that this limit potentially can be approached or even exceeded with 8 weekly fractionations without clinically relevant (grade 3+) toxicity (Fig. 4). Grade 3+ toxicities were observed for absolute lymphocyte count. Lymphocytes are recognized to be among the more radiosensitive of the blood cell types (30). Notably, lymphocytes consist of various subtypes, including B cells, T cells, and natural killer cells with different expressions of somatostatin receptors (31). A study specifically focusing on lymphocyte toxicity in PRRT reported that grade 2 or 3 toxicity was relatively common among treated patients but was considered of limited clinical relevance (32).

In a study by Sabet et al. with 203 patients treated with [177Lu]Lu-DOTATATE, the blood values took an extensive time to recover, with full recovery observed in most patients within an average of 12 mo (range, 3–22 mo) (33). A severe long-term effect of PRRT is the development of myelodysplastic syndrome, acute myeloid leukemia, and prolonged cytopenias, which occurs in approximately 5% of patients treated with [177Lu]Lu-DOTATATE (34). A study examined persistent hematologic dysfunction in a cohort of 274 gastroenteropancreatic neuroendocrine tumor patients, with the latency periods for these events ranging from 15.4 to 83.5 mo after the first treatment (34). Although absorbed dose to the RM was included as a predictive parameter, it was not found to be significant. However, it is worth noting that the study used a population-based calculation and that the absorbed dose–toxicity relationship with personalized RM absorbed dose was not conducted.

Our comparisons between absorbed doses sampled from thoracic and lumbar vertebrae demonstrated good agreement in general, although there were some outliers (Fig. 3), which may be attributed to the presence of skeletal metastases. Although efforts were made to exclude vertebrae with focal uptake, there remains a risk of misinterpreting elevated RM uptake as skeletal metastases with low uptake, and this remains a limitation. Vergnaud et al., found a significant difference between their L1–L5 and L2–4 RM representations but not other sets of vertebrae (35). Ultimately, we advocate inclusion of as many vertebrae as possible to better capture the available marrow and in case bone metastases render some vertebrae unsuitable.

A limitation of imaging-based RM dosimetry is the noise associated with the low uptake in RM. We used a numeric, analytic fitting procedure (Supplemental Fig. 1) to address this to some extent. Two particularly troublesome situations arose when the measured time–activity points either decreased to zero or remained constant for a prolonged period. These instances indicate that a noise floor was reached, rendering the activity measurements unreliable. It has been observed that the time–activity curve of the RM is, if including an uptake phase, triphasic, which makes analytic fitting an ill-posed problem with 4 imaging time points (20). Attempts have been made to address this issue, such as the use of triexponential fitting under certain assumptions (35). Considering the observed variability in shapes and the susceptibility to noise at later time points, we opted for the numeric-plus-analytic approach. Beyond the last time point, the monoexponential fit with a fixed half-life is likely to introduce a systematic overestimation in the calculated absorbed dose but not significantly impact the absorbed dose–toxicity relationship. An alternative could be using an effective half-life from the literature. Using the average value reported by Hemmingsson et al. (20), 94 h, an average percentage ± SD of −13.9 ± 5.4% was found for the cumulative activity of individual vertebrae. A limitation of our dosimetry calculation is that we assumed reference cellularity of 70% for all patients, because of the lack of patient-specific information. Patient- and region-specific CF has been explored using MRI, dual-energy CT, or sulfur colloid imaging (36–38). Unfortunately, such images were in general not available for the included patients. In the current work, we determine the BVF from the CT where there exists a possibility for beam-hardening artifacts, resulting in erroneous density estimation. Alternatives such as dual-energy or photon-counting CT (39) can mitigate the issue of beam hardening and also potentially improve tissue composition determination. To assess the impact of our assumption on the CF and the robustness of our BVF calculation, additional analysis was performed. We performed a sensitivity analysis to assess the impact of CF on absorbed dose by repeating the calculations for a subset of patients using CFs ranging from 10% to 100% and observed that the change in RM absorbed dose metrics were within ±10% if the CF ranged from 50% to 100% (Supplemental Fig. 2). Furthermore, we compared the BVF calculated for each patient and vertebra across 4 cycles and found that the relative SD was 3.2% (Supplemental Table 2).

Although the overall toxicity observed in our study did not reach clinically relevant levels, we found that blood counts remained lower than the baseline values even after 3- and 6-mo follow-up (Fig. 4) and that absorbed dose and hematologic toxicity were correlated (Fig. 5; Table 2). Although the predictive power may be limited by the variability in our data, our results are valuable to understanding and managing potential risks of PRRT, especially when considering activity escalation or individualization.

CONCLUSION

This study demonstrated a significant correlation between hematologic toxicity and absorbed dose to the RM derived from SPECT/CT imaging coupled with MC dosimetry. Although highly correlated, we observed differences using different segments of the spine to represent the RM, suggesting that, ideally, a large number of skeletal sites should be used in the RM absorbed dose calculation. Establishing an association between absorbed dose and hematologic toxicity is crucial to understanding and managing the potential hematologic risks of this therapy.

DISCLOSURE

This work was supported by grant R01CA240706 from the National Cancer Institute. Yuni Dewaraja is a consultant for MIM Software Inc. No other potential conflict of interest relevant to this article was reported.

KEY POINTS

QUESTION: Is there an association between absorbed dose to the RM, estimated using image-derived personalized dosimetry, and hematologic toxicity in patients undergoing PRRT for neuroendocrine tumors?

PERTINENT FINDINGS: This study observed a statistically significant association between absorbed doses to the RM and hematologic toxicity during and after treatment. Notably, platelet levels remained reduced at 6-mo follow-up, demonstrating the clinical significance of this correlation.

IMPLICATIONS FOR PATIENT CARE: Our findings underscore the importance of considering absorbed doses to the RM in the management of hematologic toxicity during PRRT. This personalized dosimetry approach can optimize treatment strategies, minimize adverse effects, and ultimately improve patient outcomes in PRRT.

Footnotes

  • Published online Mar. 28, 2024.

REFERENCES

  1. 1.↵
    1. Strosberg J,
    2. El-Haddad G,
    3. Wolin E,
    4. et al
    . Phase 3 trial of 177Lu-Dotatate for midgut neuroendocrine tumors. N Engl J Med. 2017;376:125–135.
    OpenUrlCrossRefPubMed
  2. 2.
    1. Ahmadi Bidakhvidi N,
    2. Goffin K,
    3. Dekervel J,
    4. et al
    . Peptide receptor radionuclide therapy targeting the somatostatin receptor: basic principles, clinical applications and optimization strategies. Cancers (Basel). 2021;14:129.
    OpenUrl
  3. 3.
    1. Lin E,
    2. Chen T,
    3. Little A,
    4. et al
    . Safety and outcomes of 177Lu-DOTATATE for neuroendocrine tumours: experience in New South Wales, Australia. Intern Med J. 2019;49:1268–1277.
    OpenUrl
  4. 4.↵
    1. Bergsma H,
    2. Konijnenberg MW,
    3. Kam BLR,
    4. et al
    . Subacute haematotoxicity after PRRT with 177Lu-DOTA-octreotate: prognostic factors, incidence and course. Eur J Nucl Med Mol Imaging. 2016;43:453–463.
    OpenUrlCrossRef
  5. 5.↵
    1. Oomen SP,
    2. van Henrik P,
    3. Antonissena C,
    4. et al
    . Somatostatin is a selective chemoattractant for primitive (CD34) hematopoietic progenitor cells. Exp Hematol. 2002;30:116–125.
  6. 6.↵
    1. Walrand S,
    2. Jamar F
    . Renal and red marrow dosimetry in peptide receptor radionuclide therapy: 20 years of history and ahead. Int J Mol Sci. 2021;22:8326.
    OpenUrl
  7. 7.↵
    1. Walrand S,
    2. Barone R
    . Experimental facts supporting a red marrow uptake due to radiometal transchelation in 90Y-DOTATOC therapy and relationship to the decrease of platelet counts. Eur J Nucl Med Mol Imaging. 2011;38:1270–1280.
    OpenUrlCrossRefPubMed
  8. 8.↵
    1. Santoro L,
    2. Mora-Ramirez E,
    3. Trauchessec D,
    4. et al
    . Implementation of patient dosimetry in the clinical practice after targeted radiotherapy using [177Lu-[DOTA0, Tyr3]-octreotate. EJNMMI Res. 2018;8:103.
  9. 9.↵
    1. Ferrer L,
    2. Kraeber-Bodéré F,
    3. Bodet-Milin C,
    4. et al
    . Three methods assessing red marrow dosimetry in lymphoma patients treated with radioimmunotherapy. Cancer. 2010;116:1093–1100.
    OpenUrlCrossRefPubMed
  10. 10.↵
    1. Wasserthal J,
    2. Breit H-C,
    3. Meyer MT,
    4. et al
    . TotalSegmentator: robust segmentation of 104 anatomic structures in CT images. Radiol Artif Intell. 2023;5:e230024.
    OpenUrl
  11. 11.↵
    1. Shah AP,
    2. Bolch WE,
    3. Rajon DA,
    4. Patton PW,
    5. Jokisch DW
    . A paired-image radiation transport model for skeletal dosimetry. J Nucl Med. 2005;46:344–353.
    OpenUrlAbstract/FREE Full Text
  12. 12.↵
    1. Wilderman SJ,
    2. Roberson PL,
    3. Bolch WE,
    4. Dewaraja YK
    . Investigation of effect of variations in bone fraction and red marrow cellularity on bone marrow dosimetry in radio-immunotherapy. Phys Med Biol. 2013;58:4717–4731.
    OpenUrlCrossRefPubMed
  13. 13.↵
    1. Hagmarker L,
    2. Svensson J,
    3. Ryden T,
    4. et al
    . Bone marrow absorbed doses and correlations with hematological response during 177 Lu-DOTATATE treatments are influenced by image-based dosimetry method and presence of skeletal metastases. J Nucl Med. 2019;60:1406–1413.
    OpenUrlAbstract/FREE Full Text
  14. 14.↵
    1. Dewaraja YK,
    2. Mirando DM,
    3. Peterson AB,
    4. et al
    . A pipeline for automated voxel dosimetry: application in patients with multi-SPECT/CT imaging after 177Lu-peptide receptor radionuclide therapy. J Nucl Med. 2022;63:1665–1672.
    OpenUrlAbstract/FREE Full Text
  15. 15.↵
    1. Isensee F,
    2. Jaeger PF,
    3. Kohl SAA,
    4. Petersen J,
    5. Maier-Hein KH
    . nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat Methods. 2021;18:203–211.
    OpenUrlCrossRefPubMed
  16. 16.↵
    1. Hough M,
    2. Johnson P,
    3. Rajon D,
    4. Jokisch D,
    5. Lee C,
    6. Bolch W
    . An image-based skeletal dosimetry model for the ICRP reference adult male: Internal electron sources. Phys Med Biol. 2011;56:2309–2346.
    OpenUrlCrossRefPubMed
  17. 17.↵
    1. O’Reilly SE,
    2. Deweese LS,
    3. Maynard MR,
    4. et al
    . An image-based skeletal dosimetry model for the ICRP reference adult female: internal electron sources. Phys Med Biol. 2016;61:8794–8824.
    OpenUrl
  18. 18.↵
    1. Peterson AB,
    2. Mirando DM,
    3. Dewaraja YK
    . Accuracy and uncertainty analysis of reduced time point imaging effect on time-integrated activity for 177Lu-DOTATATE PRRT in patients and clinically realistic simulations. EJNMMI Res. 2023;13:57.
    OpenUrl
  19. 19.↵
    1. Wilderman SJ,
    2. Dewaraja YK
    . Method for fast CT/SPECT-based 3D Monte Carlo absorbed dose computations in internal emitter therapy. IEEE Trans Nucl Sci. 2007;54:146–151.
    OpenUrlCrossRefPubMed
  20. 20.↵
    1. Hemmingsson J,
    2. Svensson J,
    3. Hallqvist A,
    4. Smits K,
    5. Johanson V,
    6. Bernhardt P
    . Specific uptake in the bone marrow causes high absorbed red marrow doses during [177Lu]Lu-DOTATATE treatment. J Nucl Med. 2023;64:1456–1462.
    OpenUrlAbstract/FREE Full Text
  21. 21.↵
    1. Menzel HG,
    2. Clement C,
    3. DeLuca P
    . ICRP publication 110. Realistic reference phantoms: an ICRP/ICRU joint effort. A report of adult reference computational phantoms. Ann ICRP. 2009;39:1–164.
    OpenUrlPubMed
  22. 22.↵
    ICRP. Basic anatomical and physiological data for use in radiological protection: the skeleton. A report of a Task Group of Committee 2 of the International Commission on Radiological Protection. Ann ICRP. 1995;25:1–80.
    OpenUrlCrossRefPubMed
  23. 23.↵
    Common Terminology Criteria for Adverse Events (CTCAE). Cancer Therapy Evaluation Program website. https://ctep.cancer.gov/protocoldevelopment/electronic_applications/ctc.htm. Updated April 19, 2021. Accessed March 8, 2024.
  24. 24.↵
    1. Strosberg J,
    2. Leeuwenkamp O,
    3. Siddiqui Mohd K
    . Peptide receptor radiotherapy re-treatment in patients with progressive neuroendocrine tumors: a systematic review and meta-analysis. Cancer Treat Rev. 2021;93:102141.
    OpenUrl
  25. 25.↵
    1. Svensson J,
    2. Hagmarker L,
    3. Magnander T,
    4. Wängberg B,
    5. Bernhardt P
    . Radiation exposure of the spleen during 177Lu-DOTATATE treatment and its correlation with haematological toxicity and spleen volume. EJNMMI Phys. 2016;3:15.
  26. 26.↵
    1. Svensson J,
    2. Berg G,
    3. Wängberg B,
    4. Larsson M,
    5. Forssell-Aronsson E,
    6. Bernhardt P
    . Renal function affects absorbed dose to the kidneys and haematological toxicity during 177Lu-DOTATATE treatment. Eur J Nucl Med Mol Imaging. 2015;42:947–955.
    OpenUrlCrossRefPubMed
  27. 27.↵
    1. Sandström M,
    2. Garske-Romań U,
    3. Granberg D,
    4. et al
    . Individualized dosimetry of kidney and bone marrow in patients undergoing 177Lu-DOTA-octreotate treatment. J Nucl Med. 2013;54:33–41.
    OpenUrlAbstract/FREE Full Text
  28. 28.↵
    1. Chicheportiche A,
    2. Ben-Haim S,
    3. Grozinsky-Glasberg S,
    4. et al
    . Dosimetry after peptide receptor radionuclide therapy: impact of reduced number of post-treatment studies on absorbed dose calculation and on patient management. EJNMMI Phys. 2020;7:5.
    OpenUrl
  29. 29.↵
    1. Benua RS,
    2. Cicale NR,
    3. Sonenberg M
    . RRW. The relation of radioiodine dosimetry to results and complications in the treatment of metastatic thyroid cancer. AJR. 1962;87:171–182.
    OpenUrl
  30. 30.↵
    1. Heylmann D,
    2. Rödel F,
    3. Kindler T,
    4. Kaina B
    . Radiation sensitivity of human and murine peripheral blood lymphocytes, stem and progenitor cells. Biochim Biophys Acta. 2014;1846:121–129.
    OpenUrlCrossRefPubMed
  31. 31.↵
    1. van Hagen PM
    . Somatostatin receptor expression in clinical immunology. Metabolism. 1996;45:86–87.
    OpenUrlCrossRefPubMed
  32. 32.↵
    1. Sierra ML,
    2. Agazzi A,
    3. Bodei L,
    4. et al
    . Lymphocytic toxicity in patients after peptide-receptor radionuclide therapy (PRRT) with 177 Lu-DOTATATE and 90 Y-DOTATOC. Cancer Biother Radiopharm. 2009;24:659–665.
    OpenUrlCrossRefPubMed
  33. 33.↵
    1. Sabet A,
    2. Ezziddin K,
    3. Pape U-F,
    4. et al
    . Long-term hematotoxicity after peptide receptor radionuclide therapy with 177Lu-octreotate. J Nucl Med. 2013;54:1857–1861.
    OpenUrlAbstract/FREE Full Text
  34. 34.↵
    1. Bergsma H,
    2. Van Lom K,
    3. Raaijmakers MHGP,
    4. et al
    . Persistent hematologic dysfunction after peptide receptor radionuclide therapy with 177Lu-DOTATATE: incidence, course, and predicting factors in patients with gastroenteropancreatic neuroendocrine tumors. J Nucl Med. 2018;59:452–458.
    OpenUrlAbstract/FREE Full Text
  35. 35.↵
    1. Vergnaud L,
    2. Giraudet A-L,
    3. Moreau A,
    4. et al
    . Patient-specific dosimetry adapted to variable number of SPECT/CT time-points per cycle for 177Lu-DOTATATE therapy. EJNMMI Phys. 2022;9:37.
    OpenUrl
  36. 36.↵
    1. Salas-Ramirez M,
    2. Tran-Gia J,
    3. Kesenheimer C,
    4. et al
    . Quantification of fat fraction in lumbar vertebrae: correlation with age and implications for bone marrow dosimetry in molecular radiotherapy. Phys Med Biol. 2018;63:025029.
    OpenUrl
  37. 37.
    1. Salas-Ramirez M,
    2. Lassmann M,
    3. Tran-Gia J
    . Quantification of the volume fraction of fat, water and bone mineral in spongiosa for red marrow dosimetry in molecular radiotherapy by using a dual-energy (SPECT/)CT. Z Med Phys. 2022;32:428–437.
    OpenUrl
  38. 38.↵
    1. Peterson A,
    2. Wilderman S,
    3. Wong KK,
    4. Frey K,
    5. Dewaraja Y
    . Monte Carlo bone marrow dosimetry in radioligand therapy using patient-specific marrow and lesion localization from quantitative 99mTc-sulfur colloid and 177Lu-PSMA-617 SPECT/CT [abstract]. J Nucl Med. 2023;64(suppl 1):P162.
    OpenUrl
  39. 39.↵
    1. Danielsson M,
    2. Persson M,
    3. Sjölin M
    . Photon-counting x-ray detectors for CT. Phys Med Biol. 2021;66:03TR01.
    OpenUrlCrossRefPubMed
  • Received for publication October 11, 2023.
  • Revision received February 21, 2024.
Back to top

In this issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools
Bookmark this article

Jump to section

Keywords