Low-dose hypersensitivity: current status and possible mechanisms

  • Published on

  • View

  • Download

Embed Size (px)


<ul><li><p>PII S0360-3016(00)01471-1</p><p>ICTR 2000 Biology</p><p>LOW-DOSE HYPERSENSITIVITY: CURRENT STATUS ANDPOSSIBLE MECHANISMS</p><p>MICHAEL C. JOINER, PH.D.,* BRIAN MARPLES, PH.D.,* PHILIPPE LAMBIN, M.D., PH.D.,SUSAN C. SHORT, M.B., PH.D.,* AND INGELA TURESSON, M.D., PH.D.</p><p>*Gray Laboratory Cancer Research Trust, Mount Vernon Hospital, Northwood, Middlesex, United Kingdom; Department of RadiationOncology, University of Maastricht, Maastricht, The Netherlands; Department of Oncology, Academic Hospital, Uppsala, Sweden</p><p>Purpose: To retain cell viability, mammalian cells can increase damage repair in response to excessive radiation-induced injury. The adaptive response to small radiation doses is an example of this induced resistance and hasbeen studied for many years, particularly in human lymphocytes. This review focuses on another manifestationof actively increased resistance that is of potential interest for developing improved radiotherapy, specifically thephenomenon in which cells die from excessive sensitivity to small single doses of ionizing radiation but remainmore resistant (per unit dose) to larger single doses. In this paper, we propose possible mechanisms to explainthis phenomenon based on our data accumulated over the last decade and a review of the literature.Conclusion: Typically, most cell lines exhibit hyper-radiosensitivity (HRS) to very low radiation doses (</p></li><li><p>induced-resistance phenomena. For HRS/IRR, the argumentwas that low doses of radiation would eliminate cells pre-dominantly in sensitive phases of the cycle and that higherdoses would then be needed to kill cells in resistant phases.(This two-population explanation is similar to the situa-tion of oxic and hypoxic cell populations in tumors, whichrespond with differing sensitivity.) Joiner et al. (9) havediscounted this two-population hypothesis. For the adaptiveresponse, the argument was that a first dose of radiationwould produce partial synchrony by eliminating cells pre-dominantly in sensitive phases of the cycle. The survivingcohort(s) of cells would then re-present in resistant phasesof the cycle some time later for the second, larger radiationdose. This hypothesis has also been challenged in nonmam-malian cell systems. For example, Howard and Cowie (10)showed that in Closterium, initial doses (which were only10% of those needed to produce any measurable responseoff the shoulder) induced an almost doubling in bothincremental (increased D0) and absolute radioresistance to asubsequent larger dose. This increased radioresistance wassignificant within 1 h following the small conditioning dosebut required 6 h to reach maximum. This could not beexplained by the known variation in radiosensitivity of thesecells within the cell cycle; subsequently the same authors(11) showed that cells kept in darkness (hence in cyclearrest) demonstrated the same adaptive response and thatthis response could be inhibited by the presence of cyclo-heximide during the period between the conditioning doseand the subsequent challenge dose. Other studies reachedsimilar conclusions; for example, Horsley and Laszlo (12,13) examined synchronous cultures of Oedogonium andfound that a first dose of radiation induced resistance to asubsequent dose that was vastly in excess of any change insensitivity that could be explained on the basis of cell cycleprogression between the two doses. Bryant (14) also testedChlamydomonas, and Santier et al. (15) tested Chlorella,and the same picture emerged.</p><p>Koval (16) demonstrated HRS/IRR in the lepidopterancell line TN-368 irradiated in either air or nitrogen. Asimilar oxygen enhancement ratio was found for both thelow-dose (sensitive) component and the high-dose (moreresistant) components of survival. This survival curve sub-structure did result from genuine induction of radioresis-tance with increasing dose, but the mechanism was notdetermined (17, 18). According to Beam et al. (4), thesubstructure in these cell survival curves could only beexplainable as the sum of the individual responses of two ormore cell populations with differing radiosensitivity if thesensitivity of one of the populations was negative in thelow-dose range; in other words, survival in response toradiation would be greater than 100% for that cell popula-tion alone, which is clearly nonsensical. Based on the evi-dence for adaptive responses in photosynthesizing cells, it ismore likely that these survival curve shapes result fromdose-dependent radiosensitivity, i.e., increased or inducedradioresistance with increasing dose.</p><p>HRS/IRR in mammalian systemsImprovements in the methodology of clonogenic assays</p><p>within the last decade (1925) have also made it possible toexamine mammalian cells with sufficient accuracy to re-solve changes in radiosensitivity at doses much less than100 cGy where cell survival approaches 100%. Conven-tional colony assays cannot reliably measure radiation-pro-duced mammalian cell death in this low-dose region. Beforethese developments, similar studies were generally possibleonly in the nonmammalian systems that respond at higherdoses, as reviewed above. These methods for improving theaccuracy of cell survival measurement determine exactlythe number of cells at risk in a colony-forming assay. Thisis achieved using either a fluorescence-activated cell sorter(FACS) to plate an exact number of cells (25) or micro-scopic scanning to identify an exact number of cells afterplating (22). Using the latter technique, Marples and Joiner(26) and Marples et al. (27) were first to define HRS andIRR in the dose range less than 100 cGy in mammalian cells(V79 hamster fibroblasts). These data demonstrated HRSbelow 10 cGy, but also HRS/IRR features following X-ir-radiation but not following single-dose irradiation withhigh-LET (linear energy transfer) neutrons. As with thenonmammalian systems cited above, the low-dose X-raysubstructure was not explained by differential sensitivity ofcells in different phases of the cell cycle (26). Therefore,there is significant homology in the HRS/IRR phenomenonbetween mammalian and nonmammalian systems, leadingone to suspect that it results from a conserved stress-re-sponse mechanism.</p><p>HRS/IRR in human cellsThere are now data on the very low-dose responses of</p><p>more than 26 different human cell lines, obtained using boththe FACS and the microscopic cell location assays (2835).The HT29 line has additionally been independently con-firmed HRS positive by two different laboratories (25, 32,35), and the T98G glioma cell line has been found HRSpositive by both the FACS and microscopic cell locationassays. Therefore, low-dose hypersensitivity in human cellshas now been well documented by different laboratoriesusing different assay techniques and different conditions ofcell growth, handling, and irradiation. Figure 1 illustratesHRS/IRR in T98G human glioma cells (33). The phenom-enon is generally more pronounced in human cell lines thanin V79 cells, and the much steeper reduction in cell survivalat low doses compared with high doses is clearly visible.This comparison is indicated in Fig. 1 by the slopes as andar, which are sensitivity parameters in the induced repair(IR) mathematical equation (31) used to model these typesof data, as shown by the solid line. The linear-quadraticmodel substantially underestimates the effect of low radia-tion doses. Figure 2 summarizes the data on HRS/IRR inmammalian cell lines tested to date. These consist of colo-rectal carcinoma, bladder carcinoma, melanoma, prostatecarcinoma, cervical squamous carcinoma, lung adenocarci-noma, neuroblastoma, glioma, one nonmalignant lung epi-</p><p>380 I. J. Radiation Oncology c Biology c Physics Volume 49, Number 2, 2001</p></li><li><p>thelial line, and one primary human fibroblast line. HRS istherefore widespread. Generally, it is those cell lines mostradioresistant to 2-Gy doses that demonstrate the mostmarked low-dose HRS, but this trend is no longer significantdespite earlier evaluations on more limited data sets (31).The radiosensitivity in the HRS region of the survival curve(as) is similar for all the cell lines regardless of the extent oftheir high-dose radiation response and is generally greaterthan 1 Gy21, which is very large. Two of the cell lineslacking an HRS response are very radiosensitive at 2 Gy,but in U373 glioma (surviving fraction at 2 Gy: SF2 5 0.63)and SiHa cervix (SF2 5 0.64) lines, HRS is also undetect-able. These lines should prove good candidates for studieson mechanisms, because they can be compared with othercell lines of similar high-dose radioresistance but whichstrongly express HRS/IRR (e.g., T98G, A7, Be11, HGL21,RT112).</p><p>HRS/IRR in vivoThere is evidence that hypersensitivity in vitro translates</p><p>into additional effectiveness of fractionated radiotherapy</p><p>given in very small doses per fraction. Thus when the doseper fraction is reduced below 1 Gy, the total dose needed toproduce damage decreases in mouse skin (36), kidney (37),and lung (38). This reverse fractionation effect is pre-cisely that expected from the HRS/IRR pattern of cellsurvival following low doses in cell lines, but, importantly,also implies a rapid decay of adaptive resistance in thesemammalian systems over the period between fractions. Inthese studies, this interval was 78 h. Recent work on threehuman glioma cell lines in vitro has confirmed the rapidrecovery of HRS between successive doses as shown in Fig.3. This means that HRS could be exploitable in radiotherapyby using very many dose fractions optimally around 0.5 Gy,an approach we have termed ultrafractionation (39). Ob-taining therapeutic gain with ultrafractionation requires thatmore excess sensitivity occur in the tumor than in criticalnormal tissues. This amount of increased sensitivity at lowerdoses depends on the parameter as/ar shown in Fig. 2, butalso on the rate at which the transition from ar to as occurswith decreasing dose. The worst case normal tissue testedso far is the kidney of the mouse (37), and if this iscompared with many of the glioma lines shown in Fig. 2,the predictions are promising. For example, if T98G werethe target tumor with kidney the critical normal tissue, 141fractions of 0.5 Gy per fraction to a total of 70.5 Gy wouldbe equivalent to 117 Gy in 2-Gy fractions to the tumor and60 Gy in equivalent 2-Gy fractions to the normal tissue.This is an overall therapeutic gain of almost 200%.</p><p>Whether low-dose hypersensitivity occurs in human nor-mal tissues, and to what extent, is clearly an important issue.On the one hand, demonstration of HRS would provideproof of the principle that the phenomenon so far observedin vitro and in animal models translates to the clinic; on theother hand, if the magnitude of HRS is too high in humannormal tissues, this would argue against a therapeutic gainfrom ultrafractionation. However, despite the impressivedocumentation of the HRS/IRR phenomenon in vitro, in-cluding human cell lines, and in vivo for various normaltissues in mice, there has been no clear-cut clinical evidencepublished. Stimulated by the various therapeutic possibili-ties of hyperradiosensitivity to very low doses, a clinicalstudy on this issue is going on in collaboration between thedepartments of oncology in Gothenburg and Uppsala inSweden (I. Turesson et al., personal communication). Theresults emerging from this work provide convincing evi-dence of the existence of a reverse fractionation effect fordoses per fraction below 1 Gy in human skin.</p><p>In the study, skin biopsies of 3-mm diameter are takenbefore and regularly during the course of radiotherapy ofprostate cancer patients. Prescribed tumor dose is 35 frac-tions of 2 Gy in 7 weeks using 11 or 15 MV photons.Biopsies taken from opposed lateral fields, with 5-mm bolusapplied to the left field, and at 1.5 and 3 cm outside thelateral fields, allow assessments after skin doses of 0.07,0.20, 0.45, and 1.10 Gy per fraction. The end point is thebasal cell density (BCD) in the epidermis. The Ki-67 index</p><p>Fig. 1. Survival of asynchronous T98G human glioma cells irra-diated with 240 kVp X-rays, measured using the cell-sort protocol.Each data point represents 1012 measurements. The solid lineand dashed lines show the fits of the induced repair (IR) model andlinear-quadratic (LQ) models, respectively. At doses below 1 Gy,the LQ model, using an initial slope ar, substantially underesti-mates the effect of irradiation, and this domain is better describedby the IR model using a much steeper initial slope as.</p><p>381Low-dose radiation hypersensitivity c M. C. JOINER et al.</p></li><li><p>is also measured to serve as an indicator of the proliferationrate in the basal cell layer.</p><p>The mean dose responses for BCD after 0.45 and 1.10 Gyper fraction in 40 patients are compared in Fig. 4. Thesedata show a highly significant reverse fractionation effect,meaning a higher effectiveness per unit of dose of 0.45 Gycompared with 1.1 Gy in the depletion of basal cells in theepidermis. The ratio between the slopes of these doseresponse relationships (dose modifying factor [DMF]) is1.8.</p><p>A confounding effect of cellular repopulation on thesedoseresponse relationships might lead to misinterpretationof a reverse fractionation effect. However, during the first 3weeks of radiotherapy, the Ki-67 index is depressed belowthe value in unirradiated skin, followed by a significantincrease during the next 4 weeks, in a similar pattern forboth 0.45 and 1.1 Gy per fraction. The doseresponse slopesfor BCD established separately for these 2 periods giveDMFs of 2.1 and 1.7, respectively. This means that cellularrepopulation to some extent actually obscures the effect ofthe HRS/IRR phenomenon.</p><p>In a subgroup of 14 patients, BCD has also been assessedafter very low doses per fraction. Figure 5 shows the re-sponse in the dose per fraction range of 0.07 to 1.10 Gy,evaluated on biopsies taken after 20 fractions in 4 weeks.The doseresponse relationship reveals a region of steepdecrease in BCD and hyperradiosensitivity up to 0.20 Gyper fraction, which precedes a plateau in BCD and increasedradioresistance. If the linear-quadratic description applied,the response for doses per fraction below 1.10 Gy should lieabove the dashed line in Fig. 5. On the contrary, the BCD at0.07, 0.20, and 0.45 Gy per fraction is found to be signifi-cantly below that line. The DMF is 3.8 and 3.4 for 0.07 and</p><p>0.20 Gy per fraction, respectively. These clinical findingsare confirmation that the experimental determinations ofHRS/IRR in vitro and in vivo actually translate to the humansituation.</p><p>Clinical evidence of hyperradiosensitivity to low radia-tion doses has also been seen in a study evaluating post-therapeutic salivary gland function in relation to radiationdose, using a functional assay (P. Lambin et al., personalcommunication). Twenty-one patients with a histologicallyproven carcinoma of head and neck of any stage treatedexclusively by radiotherapy were included in the study.They had both a CT and a scintigraphy with free 99mTc-pertechnetate in the treatment position with mask, befor...</p></li></ul>


View more >