|Home | About | Journals | Submit | Contact Us | Français|
Prediction of risks and therapeutic outcome in nuclear medicine largely rely on calculation of the absorbed dose. Absorbed dose specification is complex due to the wide variety of radiations emitted, nonuniform activity distribution, biokinetics, etc. Conventional organ absorbed dose estimates assumed that radioactivity is distributed uniformly throughout the organ. However, there have been dramatic improvements in dosimetry models that reflect the substructure of organs as well as tissue elements within them. These models rely on improved nuclear medicine imaging capabilities that facilitate determination of activity within voxels that represent tissue elements of about 0.2 – 1 cm3. However, even these improved approaches assume that all cells within the tissue element receive the same dose. The tissue element may be comprised of a variety of cells having different radiosensitivities and different incorporated radioactivity. Furthermore, the extent to which nonuniform distributions of radioactivity within a small tissue element impact the absorbed dose distribution is strongly dependent on the number, type, and energy of the radiations emitted by the radionuclide. It is also necessary to know whether the dose to a given cell arises from radioactive decays within itself (self-dose) or decays in surrounding cells (cross-dose). Cellular response to self-dose can be considerably different than its response to cross-dose from the same radiopharmaceutical. Bystander effects can also play a role in the response. Evidence shows that even under conditions of “uniform” distribution of radioactivity, a combination of organ dosimetry, voxel dosimetry and dosimetry at the cellular and multicellular levels can be required to predict response.
Prediction of radiation risks and therapeutic outcome in nuclear medicine largely relies on calculation of the absorbed dose. However, accurate specification of absorbed dose is hampered by many factors, among them heterogeneous tissue composition, nonuniform distribution of radioactivity in organs and tissues, and complex clearance kinetics. Following administration of a radiopharmaceutical, radioactivity is taken up by various organs within the body and is then eliminated through both biological clearance and physical decay. To provide a consistent and organized approach to estimate absorbed doses from tissue incorporated radioactivity, a general formalism was developed by the Medical Internal Radiation Dose (MIRD) Committee of the Society of Nuclear Medicine (1–3).
The MIRD formalism has traditionally been used to calculate mean organ absorbed doses for the purpose of risk estimation in diagnostic nuclear medicine. This approach was driven, in part, by medical imaging technologies that enabled determination of organ activity. Mean organ absorbed dose estimates assume that radioactivity is distributed uniformly throughout the organ (4). Major improvements in nuclear medicine imaging technology have facilitated determination of activity within tissue elements that are represented by voxels with volumes of about 0.2 cm3. This capability made it possible to obtain absorbed dose distributions with spatial detail commensurate with voxel size (5). As a consequence, there have been dramatic improvements in MIRD dosimetry models that reflect the substructure of organs as well as tissue elements within them (6–8). Further advances in these techniques continue (9–12).
Although improvements in nuclear medicine imaging technology have enabled us to specify absorbed dose at increasingly small spatial scales, limitations on spatial resolution require one to assume that all cells within the tissue element receive essentially the same absorbed dose. However, any given tissue element that is represented by a voxel may be comprised of a variety of cells having very different amounts of incorporated radioactivity (13–15). Furthermore, the tissue element may be composed of an assortment of cells having very different radiosensitivities (16). Accordingly, the absorbed dose delivered to the various cells in the tissue element may differ markedly as well as their biological response (17–21). To address some of these issues, the MIRD Committee published a monograph that provided a first step toward simplifying calculation of absorbed dose to the cell from intracellularly localized radioactivity (22).
It is clear that a combination of organ dosimetry, voxel dosimetry and dosimetry at the cellular and multicellular levels will be required to accurately predict biological response to nonuniform distributions of radioactivity. The microdosimetry community has long recognized the importance of the microscopic distribution of radioactivity on absorbed dose specification and ultimately biological response (23). Yet, experimentally validated approaches that can predict biological responses to nonuniform distributions of radioactivity are generally not available at the cellular level. This article describes the challenges and progress in understanding the biological effects caused by tissue-incorporated radioactivity and the development of multicellular dosimetry approaches to predict the response. Finally, the potential for fusing organ, voxel and multicellular dosimetry techniques to predict biological responses in vivo are discussed.
The extent to which nonuniform distributions of radioactivity within a small tissue element impact the absorbed dose distribution, and ultimately the biological effect, is strongly dependent on the number, type, and energy of the radiations emitted by the radionuclide. Many radionuclides used in nuclear medicine decay by electron capture and/or internal conversion (e.g. 67Ga, 99mTc, 111In, 123I, 125I, 201Tl) and consequently emit a large number of low-energy Auger electrons. The majority of these electrons deposit their energy over subcellular dimensions and therefore produce nonuniform dose distributions (24, 25). Similarly, the short range of alpha particles in biological tissues (40–100 µm) also leads to nonuniform dose distributions from 222Rn and other alpha particle emitters of potential use in nuclear medicine (25–28). Energetic beta emitters such as 90Y have a greater degree of cross-irradiation because their mean range in tissue is at least several hundred µm. However, the nonuniform distribution of these radionuclides invariably leads to nonuniform dose distributions as well (5, 21, 29, 30). Finally, the beta particles emitted by 3H deposit their energy over short distances.
While it is essential to consider the dose distributions that arise from nonuniform distributions of radioactivity, it is also necessary to know whether the dose to a given cell arises from radioactive decays within itself (self-dose) or decays in surrounding cells or other parts of the body (cross-dose). Cellular response to self-dose delivered by a radiopharmaceutical can be considerably different than its response to cross-dose from the same radiopharmaceutical. This is well known for Auger emitters where the RBE of the self-dose can be an order of magnitude greater than the cross-dose (31). We have observed this for beta emitters as well where the self-dose from 131I was 3 times more lethal than the cross-dose (see below) (32).
A simple three dimensional tissue model was developed to study biological effects of nonuniform distributions of radioactivity at the multicellular level. As described in detail elsewhere (33–35), this was accomplished by assembling 7.9 mg (~ 0.0079 cm3) multicellular clusters containing 4×106 V79 Chinese hamster cells wherein 1%, 10%, 50%, or 100% of the cells were radiolabeled and randomly located in the cluster. Thus, from a voxel perspective the activity distribution might be considered uniform, however, at the multicellular level, the distribution is clearly nonuniform. The multicellular clusters were then maintained at 10.5°C for 72 h after which they were dispersed and the cells seeded for colony formation. Thus, with this model, the variables affecting distribution of radioactivity can be tightly controlled.
The 3D multicellular cluster model was used to examine the impact of nonuniformities at the multicellular level on the lethal effects of 131I (35). When 100% of the cells were labeled with 131I-iododeoxyuridine (131IdU), the surviving fraction of cells in the cluster was exponentially dependent on the cluster activity down to 0.1% survival. The lack of a shoulder in the response to this low-LET β-particle emitter is likely due to its incorporation into the DNA (32, 35, 36). When 10% of the cells were labeled, it was observed that the survival fraction began to saturate at about 1% survival. Absorbed dose estimates revealed that the mean lethal cluster doses were 4.5, 5.7, and 6.4 Gy for 100%, 10%, and 1% labeling, respectively. Based on these data, it was concluded that when the distribution of 131I is uniform at the macroscopic level, but nonuniform at the multicellular level, the mean absorbed dose to a tissue element may not be a suitable quantity for use in predicting biological effect of 131I.
Subsequent studies were carried with this model to determine the effects of 131IdU individually on the labeled and unlabeled cells (32). In these studies the labeled cells were dyed with CFDA-SE and the labeled (dyed) and unlabeled (undyed) cells were separately seeded for colony formation using a fluorescence activated cell sorter (FACS). Unlabeled cells received only a cross-dose. In contrast, the labeled cells received both a self-dose and cross-dose. Isolating the effects of the self-dose showed that it was about 3.3 times more radiotoxic per unit dose than the cross-dose. Thus, the self-dose delivered by 131IdU has a higher RBE than the cross-dose. It appears that the higher RBE for the self-dose is related to the location in which the 131I decays occur in these cells (e.g. within the DNA) (32, 35, 36). With this in mind, it is worth noting that similar RBE values have also been reported for cell killing by DNA-incorporated 3H (emits very low energy β particles) compared to tritiated water (37).
Multicellular cluster survival data have also been obtained when cells were labeled with 210Po-citrate. About 72% of this radiochemical localizes in the cytoplasm and 28% in the nucleus (38). Figure 1 shows the surviving fraction of cells in the multicellular cluster as a function of the cluster dose when 1%, 10%, 100% of the cells were radiolabeled. Calculation of the mean absorbed dose to the cluster yielded D37 values of 1.1, 0.76, and 0.70 Gy for 1%, 10%, 100% labeling with 210Po-citrate, respectively. In contrast, the D37 value for chronic 137Cs γ-rays was 12 Gy (33–35). All three 210Po survival curves, including the 100% labeling curve, exhibited saturation despite the alpha particles range of > 5 cell diameters and the fact that each cell has about 12 neighbors. As noted above, 131IdU exhibited a similar saturation, whereas, irradiation with chronic external 137Cs γ-rays produced a classic shouldered survival curve without saturation that is characteristic of low-LET radiations (35).
It is apparent that even under these “ideal” conditions, the mean absorbed dose to a very small tissue element may not be a suitable quantity for use in predicting biological effects of alpha particle emitters (e.g. 210Po) and relatively long-range beta emitters (e.g. 131I). Initial modeling of the killing of labeled and unlabeled cells was achieved by considering the cellular self- and cross-doses and the fraction of cells labeled (39).
where Dself is the mean self-dose to the labeled cell, Dcross is the mean cross dose received by the labeled and unlabeled cells, f is the fraction of cells labeled, and D37,self and D37,cross are the mean lethal self- and cross-doses, respectively. This approach was used to model the observed responses for 210Po-citrate (not shown) and 131IdU (39). The model afforded an excellent fit to the first two logs of killing in all three labeling cases (1%, 10%, 100%). This shows the promise of using multicellular dosimetry models to predict response. However, this model was not successful at predicting the response beyond the second log of killing due to the saturation in the experimental survival curves.
In the cases of 210Po and 131I, survival curves for 10% labeling exhibited saturation despite the high degree of cross-irradiation afforded under these conditions by both 131I and 210Po (Fig. 1). The survival curve for 100% labeling with 210Po also saturated to some degree (Fig. 1). These are important findings in that they have significant consequences for therapeutic use and risk estimation for these and similar radionuclides. We hypothesize that this saturation is related to a combination of variations in uptake of radioactivity by labeled cells and the geometry or shape of the cluster. Geometry can have a significant impact on dose distribution. Even when all cells in a spherical cluster contain the same 131I activity, cells at the center of a spherical cluster receive double the cross-dose received by those at the periphery. Measurements of our multicellular clusters show that the geometry of the cluster is a paraboloid capped with a cylindrical wedge. This geometry, coupled with nonuniformities in uptake of radioactivity (see below), can lead to an even wider range of cellular doses. Autoradiographic studies which quantify the relative cellular uptake of 210Po-citrate have revealed a log normal distribution of radioactivity per labeled cell (40). This broad log normal distribution has emerged even though all of the cells in a uniform suspension were exposed to the same extracellular concentration of 210Po-citrate. A similar distribution was observed for 131IdU. Initial modeling calculations show that this activity distribution can result in survival curves that saturate in a manner analogous to those observed in Fig. 1 (40). More detailed Monte Carlo modeling is under way with the ultimate goal of matching the observed survival curves.
Recent developments in radiobiology have shown that irradiated cells can prompt biological responses in neighboring unirradiated cells. These phenomena are known as bystander effects. Numerous studies on bystander effects have been observed following external beam irradiation of a variety of cell types in a number of cell culture models, as well as in vivo models (41). Observed bystander effects include genetic alterations (mutations, micronucleus formation), neoplastic transformation, changes in gene expression and cellular redox status, induction of apoptosis, alteration of proliferation, and modulation of cell cycle check points (42). Furthermore, genomic instability is thought to be mediated via bystander mechanisms (43). Bystander effects have been shown to be mediated by gap-junction intercellular (GJIC) communication (41, 44, 45) and by signaling via diffusible secreted factors (46, 47). Bystander effects can be modulated by oxidative metabolism (41, 48, 49). These findings and their implications have been extensively discussed in a number of review articles that address the bystander effect (50, 51).
These findings suggest that bystander effects are highly relevant to the biological effects of nonuniform distributions of radioactivity. This is borne out by our experimental data in three dimensional cell cultures held at 10.5°C that demonstrates bystander cell killing in unlabeled cells adjacent to cells labeled with 3HTdR or 125IdU (33, 34, 52). Using a similar model, Persaud et al. have observed bystander cell killing when 20% of the cells were labeled with 3HTdR (53). In addition, they showed that the 3HTdR labeled cells imparted mutagenic effects on their neighbors. Xue et al. used an in vivo tumor model to show that labeling cells with 125IdU led to an antiproliferative response in the bystander cells (54). Surprisingly, when cells were labeled with 123IdU, an Auger electron emitter with similar properties to 125I, a proliferative bystander response was observed (A.I. Kassis, data presented at this meeting – Microdosimetry 2005).
Proliferative bystander responses have also been observed in monolayer cultures of unlabeled cells adjacent to cells labeled with 3HTdR (55). This response was shown to depend on the percentage of cells labeled, with ≥50% being required (56). In this model, the proliferative bystander response did not appear to be mediated by either gap junctional intercellular communication or long range factors released from the irradiated cells (57).
Additional bystander studies have been carried out in our laboratory with three dimensional cultures of apparently normal human fibroblasts growing in Cytomatrix™ carbon scaffolds. In these studies, cells were co-pulse-labeled in situ with 3H-deoxycytidine and bromodeoxyuridine (BrdU). Using this labeling approach, only cells in S-phase were labeled (3–16% of total population) and the labeled cells were identified using standard anti-BrdU flow cytometric approaches. Induction of a G1 checkpoint in bystander cells was assayed by monitoring the transition from G1 to S phase using uptake of iododeoxyuridine (IdU) and a three-parameter flow cytometry technique. It was determined that dose rates to the labeled cells up to 0.32 Gy/h did not induce any significant G1 checkpoint in the unlabeled bystander cells. It is possible that no significant stress-induced effects were observed because the dose rates were insufficient and/or the percentage of cells labeled was not high enough to elicit a bystander response. Details of these studies are presented in a companion manuscript in this issue of Radiation Protection Dosimetry (Pinto et al.).
It is clear that a variety of bystander responses can be elicited by radiolabeled cells. The type and magnitude of the response depend on the radionuclide, percentage of cells labeled, self-absorbed dose to the cells, and tissue microenvironment. It is clear that modeling the biological response of cells to nonuniform distributions of radioactivity solely on the basis of absorbed dose may not adequately predict response even when the dose is calculated at the cellular level. This is of importance to risk estimation in diagnostic nuclear medicine and radiation protection (radon, radiological terrorism), as well as clinical outcome in therapeutic nuclear medicine.
Prediction of biological responses to incorporated radioactivity will require a comprehensive dosimetry approach. Ultimately, knowledge of absorbed dose at the cellular level is essential. The dose to any given cell is delivered by radiations emitted by radioactivity in the given cell (self-dose) as well as radioactivity in other locations in the body (cross-dose). Cross-doses are delivered by decays in organs afar and in nearby tissue elements (voxels). These cross-doses can be obtained with the aid of modern nuclear medicine imaging techniques. Additional cross-dose arises from decays in neighboring cells within the same tissue element. Multicellular approaches such as those described herein are required for this contribution. The cellular self-dose can be calculated using standard cellular S value approaches. Careful bookkeeping is required to keep track of the self- and cross-doses from each type of radiation so that their respective RBE values can be applied. Finally, multicellular dose response modeling will be required to account for these variables as well as those relating to bystander effects.