Mucosal regeneration during radiotherapy

ELSEVIER

Radiotherapy and Oncology 41 (1996) 109-l 18

Mucosal regeneration during radiotherapy ’ James W. Denham”‘*, Quenten J. Walkerb, David S. Lamb”, Christopher S. Hamilton”, Peter C. O’Brien”, Nigel A. Spryd, Andrew Hindley”, Michael Poulsenb, Maree O’Brien”, Lee Tripconyb “Department

of Radiation

Oncology,

Newcastle

Mater

Misericordiae Hospital, Locked Bag 7, Hunter Region Mail NSW 2310, Australia ‘Queensland Radium Institute, Royal Brisbane and Mater Hospitals, Brisbane, QLD, Australia ‘Wellington Regional Oncology Unit, Wellington Hospital, Wellington, New Zealand ‘Radiation Oncology, Andrew Love Centre, Geelong Hospital, Geelong, VIC, Australia ‘Clinical Oncology, Auckland Hospital, Auckland, New Zealand

Centre,

Waratah,

Received 16 April 1996; revised 18 June 1996; accepted 30 June 1996

Abstract Background and purpose: Regeneration of the aerodigestive mucosa is known to occur during conventionally fractionated radiotherapy. The circumstances surrounding its time of onset and magnitude are not well understood, however. Material and methods: Mucosal reactions were observed in 100 patients undergoing conventionally fractionated treatment at 2 Gylday over 7 weeks and 88 receiving accelerated treatment at 1.8 Gy twice daily over 3f weeks on the Trans Tasman Radiation Oncology Group head and neck cancer trials. Similar observations in 61 patients treated palliatively at dose rates between 0.8 and 240 Gy/h using ten 3.0-4.2 Gy fractions over 2 weeks are compared. Results: Several findings emerged from these studies: 1. Reactions evolved more quickly at oropharyngeal sites than in the hypopharynx. 2. Reactions at both sites evolved more rapidly at greater rates of dose accumulation. 3. The timing of reactions suggested the presence of a strong regenerative mucosal response that started before the manifestation of ‘patchy’ (grade II) mucosal reactions. 4. The regenerative response was strong enough to ‘make good’ damage accumulated at a rate of 2 Gy/day in over a third of cases. 5. The linear quadratic model without time correction failed to provide an adequate prediction of the frequency or intensity of mucosal reactions produced by any of the regimes. A simple model of the regenerative response is presented. Conclusions: This study suggests that the timing and magnitude of the regenerative response vary between sites and individuals but are linked to the amount of epithelial cellular depletion occurring during treatment. Keywords:

Head and neck cancer; Acute mucosal radiation reactions; Mucosal regeneration;

1. Introduction Current interest in acceleratedfractionation regimes for the treatment of head and neck cancer has drawn attention to the importance of acute mucosal reactions. Unfortunately the present state of knowledge of these reactions is not as detailed as it could be. It is known, for example, that a variation in the time course of mucosal reactions at different head and neck sites exists [16] but precise quantification of these differences for any given fractionation scheduledoes not exist. It is also known that mucosal regeneration starts during a conventionally frac-

Fractionation

studies

tionated course of radiation and can ‘make good’ the damagecausedby up to 1.8 Gy per day [15]. However, it is not known when the regenerative responsecommences nor how much mucosal cell kill it can make up for at any particular site for any particular fractionation schedule.An elegant series of experiments on the mouse mucosa by Diirr and coworkers [4] has suggestedmechanismsthat could produce a massive regenerative response,but it is unknown whether such mechanisms are operative in humans. Smoking [13] and bacterial infection [9] have been identified as factors that can influence the severity and duration of mucosalreaction in humans,but the degree of interpatient variation that exists in the absenceof these factors for any particular fractionation schedule has not been thoroughly documented.Finally, it is not certain how

*Corresponding author. Tel: +61 49 211177; fax: +61 49 602566. ‘On behalf of the Trans Tasman Radiation Oncology Group (TROG). 0167-8140/96/$15.00 0 1996 Elsevier Science Ireland Ltd. All rights reserved PZZ SO167-8140(96)01830-O

110

J.W. Denham

et al. I Radiotherapy

endpoints that are readily observed in the clinic, such as the patchy and confluent ‘mucositis’ reactions that are used in most reaction grading scales, relate in any quantitative fashion to mucosal germinal cell killing and cellular density. The present report draws on a detailed set of clinical observations of the reactions that took place in patients undergoing a range of different fractionation schedules to shed light on some of these unknowns.

and Oncology

41 (1996)

109-118

institution phase III trial referred to above between 1992 and 1995 at centres other than the Mater Hospital, Newcastle. The treatment centres involved are the Queensland Radium Institute: Royal Brisbane and Mater Hospital centres (Queensland, Australia), Wellington Hospital (New Zealand), Christchurch Hospital (New Zealand), Prince of Wales Hospital (Sydney, Australia), Geelong Hospital (Victoria, Australia), Auckland Hospital (New Zealand) and Waikato Hospital (New Zealand). Approval for participation in the trial was granted by the respective regional ethics committees.

2. Patients and methods 2.2. Reaction scoring 2.1. Patients The patients reported have come from three studies: 2.1.1. ‘Low dose rate teletherapy study’ patients Sixty one patients with incurable head and neck cancer were treated on this study between 1988 and 1993 to determine the radiobiological parameters (alp, t+) for acute mucosal reactions. Details have been reported previously [3]. Briefly, 51 patients were treated at various low and intermediate midline dose rates (0.8, 1.8 and 3.0 Gylh) using opposing lateral telecaesium beams. Total midline doses between 32 and 42 Gy were delivered using 10 daily fractions. Ten additional patients were treated at conventional high dose rate (240 Gy/h) using opposing lateral 6 MV linear accelerator fields to 30 Gy midline in 10 daily fractions. The study was approved by the Hunter Area Ethics Committee in 1988.

Acute mucosal reactions have been scored according to the EORTC/RTOG grading scale which is reproduced in Table 1. Reactions in patients treated at the Mater Hospital, Newcastle, on the ‘low dose rate’ and ‘fractionation’ studies have, where possible, been scored on a twice weekly basis by two experienced observers during treatment. For patients treated on the ‘low dose rate’ study, twice weekly scoring continued until the resolution of reactions, but for patients treated on the ‘fractionation’ study, scoring was carried out where possible on a weekly basis after the completion of treatment. Fibreoptic equipment has been used to assist accurate scoring of reactions in the hypopharynx and larynx. In some of the patients, full sets of reaction grades from commencement of treatment to full reaction healing is not available and reaction recovery profiles are therefore not reported herein. For patients treated on the ‘phase III trial’, at centres other than the Mater Hospital, Newcastle, reaction scoring by the treating practitioner was on a weekly basis during treatment and as frequently as circumstances allowed following treatment, but at least monthly.

2.1.2. Mater Hospital, Newcastle, conventional and accelerated ffractionation study ’ patients Sixty-nine patients with potentially curable head and neck cancer between 1990 and 1995 were treated on the Trans Tasman Radiation Oncology Group (TROG) phase II and III studies. Details of the accelerated regime that formed the basis for the TROG multicentre phase II study conducted in 1990 and 1991 have been reported previously [lo]. The phase III study, in which patients with Stage III and IV cancer were randomised to conventionally fractionated treatment to 70 Gy in 35 daily 2 Gy fractions over 7 weeks or to accelerated fractionated treatment to 59.4 Gy in 33 twice daily 1.8 Gy fractions with a minimum interval of 6 h over 24 days, was activated in 1991 when it became obvious from the phase II study that the accelerated regime could be ‘tolerated’ on a multi-institution basis. Participation in these studies was approved by the Hunter Area Ethics Committee in 1990 and 1991.

Data from all three studies have been entered on to a temporal database at Newcastle (MEDLOG). This database permits a range of statistical manoeuvres and enables data to be ‘downloaded’ for complex statistical manipulations. Non parametric testing has been used extensively in the analyses. The Fisher Exact and Chi-squared tests have

2.1.3. Non-Mater, Newcastle, ‘TROG phase III trial’ patients The third group is formed by 119 patients with cancers of the oral cavity and oropharynx who have been treated on the Trans Tasman Radiation Oncology Group multi-

III

2.3. Data analysis

Table 1 EORTC/RTOG Mucous 0 I II

IV

mucosal

reaction

scoring

criteria

membrane No change over base line Injection; mild pain not requiring analgesics Patchy mucositis which may produce an inflammatory serosanguinous discharge; may experience moderate pain requiring analgesics Confluent fibrinous mucositis; may include severe pain requiring narcotics Ulceration, haemorrhage or necrosis

J.W. Denham

et al. I Radiotherapy

and Oncology

been used to compare proportions and the Kruskal-Wallis to compare data distributions between groups where normality can not be assumed. Dale’s approach [2] to the incomplete repair model of Thames and colleagues [14] has been used to derive ERD,, values for the regimes used, and a custom designed spreadsheet program on Lotus has been used to model time course data (see appendices).

41 (1996)

109-118

111

onset of both grades of reaction during conventionally fractionated treatment and during the accelerated protocol regardless of site (Table 2, comparisons labelled with suffixes 5-8; all P values CO.001; and Fig. 1). The time of onset of grade II reactions in patients receiving accelerated treatment was very similar to patients treated on ‘low dose rate’ protocols using similar RDAs (Table 2 and Fig. 2).

3. Results 3.1. Time course variations 3.1.1. Site difSerences

The onset of grade II (i.e. patchy) oropharyngeal mucosal reactions were significantly earlier than hypopharyngeal reactions, regardless of the rate of dose accumulation (RDA) (Table 2, comparisons labelled with suffixes 1 and 2; P values 0.008 and
The rate of onset of grade II and III reactions was observed to bear an inverse relationship with the rate of dose accumulation whether this concept is expressed in uncorrected Gy/day or in ERD,,/day. Statistically significant differences were observed to exist between the time of Table 2 A comparison

of the onset times of grade II and III reactions

Conventional (RDA= 1.79 ERD,, Gropharynx

units/day”

Hypopharynx Accelerated (RDA=2.98 ERD,, units/day” Oropharynx fWopharynx Low dose rate (Oropharytu) RDA=2.71-2.82 ERD,,/day= RDA=2.94-3.07 ERD,,,/daysd RDA=3.09-3.14 ERD,,/day”’

in mater

patients

The rate of evolution of patchy into confluent reactions was also strongly dependent on RDA (Table 3). Grade II reactions in patients treated using conventional fractionation took significantly longer to evolve into grade III reactions than the corresponding reactions in patients treated using accelerated fractionation. This phenomenon is depicted in Figs. 1 and 2. 3.1.3. DifSerences according to total dose accumulated (TDA)

The proportion of patients who developed grade Ill reactions was observed to depend both on the rate of dose accumulation and the total dose accumulated. This phenomenon was most clearly seen in patients treated at high RDA on the ‘low dose rate’ protocols to different total accumulated doses (Table 3). Duration of reaction at both grade II and III levels was greater in patients receiving larger total accumulated doses too (Fig. 3). 3.1.4. Non-Mater

trial patients

In spite of quantitative differences, strong qualitative confirmation of the three major findings made in Mater patients came from patients treated on the TROG trial at

treated

on the ‘fractionation’

and ‘low

dose rate’ studies

No. of patients

Time of reaction Grade II

onset in days [median

(range)] Grade III

27 17

20 (S-W)‘*~ 28 (16-~+~

44 (1%@$.’ (22-w)‘.8

13 12

13 (8-15)= 15 (10-20)=

17 ( 13-28)4.7 20 ( 15-27)4.8

23 19 19

14 (6-00)~ 13 (9-m) 12 (7-16)9

22 (8-m)” 18 (11-m) 16 (9-q”

or 1.49 Gy/dayb)

or 2X 1.24 Gy/dayb)

Numeric suffixes refer to Kruskal-Wallis comparisons between pairs of groups. E.g. Suffix 1 applies to the comparison between onset of reactions at Grade II level for the oropharyngeal and hypopharyngeal groups treated with conventional fractionation. The p value for this comparison is 0.008. Suffix 2 refers to the comparison between the oropharyngeal groups treated with accelerated radiation. The p value is CO.001. The p values associated with these and the remaining comparisons are:‘P=O.O08 ‘P=O.27 5P<0.001 ‘P
112

J.W. Denhum

et al. I Radiotherapy

and Oncology

41 (1996)

Gropharynx (Mater patients)

j

Hypopharynx (Mater patients)

g

II + -a--

Accelerated

conv~uonsl

109-118

II

(ni13) (ll.27)

s o-W--

Accalamted Convmilofml

(re12) (nd7)

Proportion nrhing grad* Aacdmmad 100%

0

10

20

30 Days

40 after

50

60

70

0

10

20

30

Days after

stattlng

40

50

60

II

70

starting

Fig. 1. The timing of onset of grade II and grade III reactions in Mater Hospital patients treated on the ‘conventional’ and ‘accelerated’ TROG trial protocols. (a) presents data from patients with oropharyngeal turnours where orophaqngeal reactions have been scored and (h) from patients with hypopharyngeal and laryngeal tomours where hypopharyngeal reactions have been scored. The data points are median values and the bar represents the range of recorded values (+ m indicates that a reaction did not reach the specified grade).

other centres Fig. 4 and Table 4). Grade II reactions were significantly earlier in patients receiving accelerated treatment. In addition, these reactions evolved significantly more frequently and more quickly into grade III reactions in these patients. Quantitative differences between Mater and non-Mater data can be assessed by comparing the data in Tables 2 and 4 and may be traced back to smaller degrees of inter-observer variation at the Mater (where there is a special interest in mucosal reaction scoring) than at all other centres combined. A lesser, but not negligible, 50

i 1

0 0

0

Grade11 Grade Ill

/

I

I

0

1

2

1

3

4

RDA (ERD units/day) Fig. 2. Time of onset of grade II and III oropharyngeal reactions according to rate of dose accumulation. Data from patients treated at the ‘accumulated’ and ‘low dose rate’ protocols Mater with ‘conventional’, have been plotted. (Points are median values and bars the range).

contribution to the differences seen was the frequency of observations made. 3.2. Interpatient variation

The size of interpatient variations in time of onset of reactions at both grade II and III level was observed to reduce as RDA increased. Providing that total dose accumulated was sufficient to cause a reaction at grade II level, the distribution of times to onset of grade II reactions was approximately normal. To investigate the possibility that interpatient variation in the expression of reactions may be related to interpatient variations in normal tissue radiosensitivity or in regenerative capacity, subdivision of patients treated in each site and RDA category was carried out on the basis of the rapidity of onset of grade II reactions. It would be expected that the most rapid responders included some of the most ‘radiosensitive’ patients, whose patchy (grade II) reactions would have been more likely to evolve more frequently and rapidly into confluent (grade III) reactions. In Table 5 it will be noted that patients in the accelerated group with the most rapidly developing grade II reactions experienced grade Ill reactions no more frequently or quickly than patients who developed grade II reactions near or at the median time for the whole patient group. This observation was also true for patients whose grade II reactions evolved no more slowly than the ‘median’ group too. However, amongst patients where the RDA (rate of dose accumulation) was low (i.e. the conventionally fractionated group) or where TDA (total accumulated dose) was too low to produce grade III reactions in all cases (i.e. the ‘low dose rate’ group), those whose grade II reactions developed

J.W. Denham Table 3 A comparison

of the evolution

et al. 1 Radiotherapy

of grade II to grade III reactions

in Mater

Proportion Conventional Oropharyrix Hypopharym Accelerated Oropharynx Wpopharynx Low dose rate (Oropharynx) TDA = 37.9-39.5 ERD,, units TDA=41.2-43 EFCD,, units TDA=43.2-44 ERD,, units

(63%)1.5 (47%)1.6

13/13 12/12

(100%)3.5 (100%)3,6

Grade

5 (2-2o)+’ 4.5 (2-8)4,* 6 (2-m)” 5 (2-m) 5 (3-m)” [median

(range)] Grade

‘2.94-3.07 3.093.14 to +to --

ERDlday ERD/day

(n=lQ) (n=lQ)

III

10

20 Time

30

40

(days)

Fig. 3. Onset and recovery of reactions in three groupings of patients treated on the ‘low dose rate’ project at the Mater. Data points are median values and the bars represent the range of recorded values ( +a, and -00 values indicate that a reaction did not reach a given grade).

III

0.5 (0-9)12 4 (O-10) 4 (o-1o)‘2

refer to Kruskal-WalIis comparisons between groups (using the same labelling convention ?‘=O.S; ‘P=O.O3; ‘?‘=0.002; ‘P
RDA = RDA = Tends Tends

grade II and III

21 (3-m)2’7 27 (3-m)‘,’

8.5 (O-18)” 12.5 (O-17) 14 (1 l-29)”

The first important point to come out of these analyses was the failure of the linear quadratic model without time

113

Time in days between [median (range)]

II

4. Discussion

0

grade II

in days of reactions

more slowly than the ‘rapid’ or ‘median’ response groups experienced grade III reactions slightly less frequently and more slowly. However, these differences did not approach statistical significance.

t -c Ge

109-118

12/239 (52%) 13/19 (68%) 14/199 (74%)

Low dose rate TDA = 37.9-39.5 ERD,, units TDA =41.2-43 ERD,, units TDA=43.2-44 ERD,, units suffixes 3P=1.0;

41 (1996)

patients

reaching

17/27 8/17

Duration

Numeric ?‘=0.9;

and Oncology

as in Table “P=O.l.

3). The P values are: ‘P =O.l;

correction to provide a useful prediction of which arm of the fractionation trial would experience greater acute mucosal reactions. ERD value, uncorrected by any time factor, for the conventional fractionation schedule is 84 units if an o/p ratio of 10 is assumed. For the accelerated regime it is only 71. However, our results clearly indicate that mucositis was more frequent and severe in patients treated on the accelerated schedule. While this finding was of no surprise to us, it emphasised the fact that at present, no satisfactory way of predicting the acute mucosal reaction outcome of a fractionation trial of this type from radiobiological first principles exists. This is because reliable methods of correcting for overall time have not been worked out yet. For the designers of randomised controlled clinical trials, therefore, there remains no substitute for adequate pilot data. There are several problems involved in deriving methods of correcting for overall time. Firstly, there is the problem of knowing when the regenerative response to radiation in the mucosa starts. It is not a process that can be seen with the naked eye. Its presence may be inferred when a patient’s reaction does not worsen, or even improves during a course of radiation. In fact, it was observations of this type that lead Fletcher to postulate that mucosal regeneration does occur during conventionally fractionated radiotherapy [15]. To make predictions more difficult, it is a process that commences an unknown time before its presence can be inferred. Secondly, there is the problem of knowing what amount of damage in ‘radiation dose equivalent’ terms is ‘made good’ by regeneration during radiotherapy. Fletcher and others have suggested that the regenerative process may make good damage inflicted at a rate of 1.8 Gy/day in some patients. This is

J.W. Denham

114

Grade Grade

et al. I Radiotherapy

II Aoooleraled II Conventional

and Oncology

41 (1996)

109-118

1

1143 n&6

0 0

Grade Grade

III Aoodereled Ill Conventional

n&J n&6

0.7 .g

0.6

=3

0.5

2 n

0.4

-p D a n 2 P

0.3

0.6 0.5 0.4 0.3

0

50

100

150

200

0

50

Time in Days

loo Tim

150

200

in Days

Fig. 4. The proportion of non-Mater patients who experienced reactions at least as severe as the grade II level (a) and the grade III level (b) as a function time after starting treatment. The data points represent mean values and the bars standard errors.

because they have noted that reactions often do not evolve, and sometimes improve, after the first three or four weeks of a conventionally fractionated course of radiation. This, of course, leads to the last set of problems. These are that the timing and degree of response may vary according to the rate at which radiation damage is accumulated, from anatomical site to site and from individual to individual. The data reported herein offer some indication of the sizes of these unknown quantities. The way in which rate of evolution of reaction varies with rate of dose accumulation offers the best insight into when the regenerative response occurs. Time to the appearance of a patchy (grade II) mucositis is shorter for the accelerated regime (whose rate of dose accumulation [RDA] is 2.95 ERD,, units/day) than for the conventional regime (whose RDA is 1.79 ERD,, units/day). However the difference, although highly statistically significant, was only 7 days for both oropharyngeal and hypopharyngeal reactions in Mater patients (Table 2). It was after this point that the differences in reaction evolution between the two regimes became massive. Within a few days, reactions in virtually all patients receiving accelerated treatment proceeded from a patchy to confluent mucositis. However, in over a third of patients treated conventionally, patchy reactions did not evolve at all and the rate of evolution was highly variable and was often more than three weeks (Table 3). This Table 4 A comparison

of the onset times

of reactions

Time of reaction [median (range)] Grade

Conventional Oropharynx Accelerated Oropharynx Numeric ‘P
suffixes refer *P
observation, which is very similar to one made at Nijmegen [8], suggests that the mucosal regenerative response commences before the appearance of a patchy mucositis in all conventionally treated patients (i.e. before 20 days for the oropharynx and 22 days for the hypopharynx) and then makes good damage at a rate of 2 Gy/day (1.79 ERD,, units/day) in at least a third of cases. In the remaining patients, the regenerative process may not reach this magnitude, but is clearly sufficient to retard the evolution of reactions significantly in the great majority. Obviously, the question arises as to whether the regenerative process could also be starting before the appearance of patchy reactions in patients treated on the accelerated regime too. We believe that it does. We can think of no other way of explaining why the reactions seen in patients treated on the low dose rate protocols (which involved similar RDAs, but to lower total accumulated doses) recovered so rapidly (Fig. 3). If we are correct, the implications of this are that the commencement of a regenerative response must, under certain treatment conditions, be possible less than 13 days after starting treatment in the oropharynx and 15 days in the hypopharynx. Additional implications are that the regenerative process can not reach a magnitude sufficient to make good damage inflicted at a rate of 2.98 ERD,, units/day. If it could, then reactions would not evolve from a patchy to a

at grade II and III levels in the oropharynx

No. of patients

II

in non-mater

onset in days Grade

patients

treated

on the randomised Proportion

III

16 (6-55)’

32 (14-m)’

46/56

63

14 (S-24)’

20 (s-my

62163 (92.4%)3

comparisons

between

groups:

TROG

reaching

56

to Km&al-Wallis 3P=o.oo6.

of

(82.1%)’

trial

grade III

J.W. Denham

et al. I Radiotherapy

and Oncology

41 (1996)

115

109-118

Table 5 The evolution of grade II to grade III reactions according to the rapidity of onset of grade II reactions (Mater patients) Time between onset of grade II and III reactions [median (range)] and proportion reaching grade III Rapid responders* Conventional

Median responders*

Slow responders*

14 (3-m)‘,Z

11115

21 (3-m)’

13/17

m (14--m)*

l/9

5 (3-20)

919

4 (2-7)

717

5.5 (2-7)

8/8

5 (1-m)3

13/18

4 (2-m)

17124

cc (2-m)3

7115

Oro and hypopharynx Accelerated

Oro and hypopharynx Low dose rate

Oropharynx Numeric suffixes refer to Kruskal-Wallis comparisons between groups: ‘P=O.9; ‘P=O.48; ‘P=O.49. * ‘Rapid responders’ are patients whose grade II reactions manifest amongst the earliest (33.3%) of cases in a particular group, ‘median responders’ are those whose grade II reactions manifest amongst the next 33.4-66.6%, and ‘slow responders’ are those whose grade II reactions manifest latest amongst the final 66.7-100%.

confluent state in some patients. However, lesser magnitudes of responseare clearly possible, becausealthough reactions evolve to the confluent stage in less than 5 days in half of all patients, evolution is sometimesconsiderably slower in the remaining half (Table 3). Readerswill undoubtedly be speculating whether variations in radiosensitivity, mucosal germinal cellular density, and natural rates of exfoliation are responsiblefor someof the variations in reaction evolution that we have observed. Of course, it is quite possible that any or all of these parameterscould be responsiblefor the significant differences in time course noted between oropharyngeal and hypopharyngeal reactions. It is widely held that tumours arising in the hypopharynx are more radioresistant than tumours of the soft palate and tonsil. It would not be surprising, therefore, if the mucosa from which these tumours originated differed in sensitivity in the sameway. If differences in radiosensitivity are responsiblefor site to site differences observed, then it would be expected that the more radiosensitive tissue would be first to express damagein the form of a visible reaction. This would mean that our observations suggestthat oropharyngeal mucosais more sensitive than the mucosa of the hypopharynx. Although this conclusion is plausible, it must be pointed out that differences in overall mucosal cellular density at different head and neck sites have been documented in rodents and humans [5]. Since labelling indices at the various sites were found to be similar, it is possible that differences in cellular turnover time could also be responsible for the differences in reaction expression time between sites [5]. Obviously, the question also arises as to whether interpatient variations in radiosensitivity and/or mucosal cellular density might explain the substantial interpatient differences in time to expression of reaction grade. If variations in radiosensitivity were responsible for the interpatient variations observed, then it would, of course, be expected that the most ‘radiosensitive patients’ would exhibit more intense reactions and that these reactions would be manifest earlier. We have attempted to address

this issue with the analysis presented in Table 5. In this analysis, patients in each site/fractionation/T group have been divided into three equal subgroupsaccording to time of expression of a patchy mucositis. The subgroup comprising patients whose reactions evolved most quickly is likely to contain patients whose mucosaeare among the most radiosensitive and whose reactions might be expected to evolve from the patchy to confluent states more frequently and more quickly. There is little indication from Table 5, however, that this is the case.We therefore favour the conclusion that, while interpatient variation in radiosensitivity and/or mucosal cellular density and/or natural rates of exfoliation may be responsiblefor variation in the timing of ‘patchy’ grade II reactions, interpatient variations in the timing and magnitude of the regenerative responseare more important contributors to the variation in evolution of reactions from the grade II to III levels in this study. Van der Schueren and colleagueshave concluded from observations of mucosal reactions in patients undergoing a variety of multiple daily fraction pilot protocols, that confluent mucositis will become manifest 8 or 9 days after a threshold dose of 20 Gy has been delivered (quite regardlessof the RDA) [ 181. This suggestionis appealing becauseit embodiesthe notion that a given reaction grade correspondsto a certain level of cellular depletion caused by the accumulation of a certain amount of radiation damage.It also embodiesthe idea that an interval of time will passbetween the time at which damage sufficient to cause a given grade of reaction has accumulated and the time at which the reaction becomes manifest. To investigate the possibility that the regenerative response is ‘triggered’ by the accumulation of a given amount of radiation damage (i.e. a given level of basal cellular depletion in the proliferative compartment), we have constructed a simple model of the time course of reactions experienced in our fractionation study assumingexponential regeneration, as suggestedby Liversage [l l] and later elaborated by Fowler [7] and Barendsen [l] (see Appendix). This approach involves the notion that at a certain

116

J.W. Denham

et al. I Radiotherapy

point in time during radiotherapy, regeneration commences and then proceeds exponentially. In the model that we have derived for oropharyngeal reactions in Mater patients, which is depicted in Fig. 5, we have made the additional assumption that regenerative responses of identical magnitude are triggered when identical levels of damage (expressed in ERD units) are accumulated by both the conventional and accelerated regimes. It will be noted that the model, which assumes that a regenerative response equivalent to 1.56 ERD,, units/day occurs when sufficient ERD units have been delivered during both regimes to cause a grade II mucositis, is successful in predicting the median timing of the expression of grade III mucositis (Table 2) for both regimes. Less easy to see in the figure is the fact that the model incorporates a further increase in the regenerative response to 1.79 ERD,, units/day when 38 ERD,, units of damage have been accumulated. This additional increase successfully predicts the clinical finding that reactions in almost one half of patients treated on the conventional regime do not become confluent. Support for the concept of an increase in regenerative response has come from cell density studies both in the pig skin [12] and in mouse tongue mucosa [6]. Both of these studies indicate that greater rates of regeneration are associated with radiation doses that produce lower levels of stem cell depletion. Obviously the use of simple models to explain complex biological processes should be regarded with the utmost scepticism. This particular model is interesting because it Time (days)

:IIIr

g* -3oIll grade III (confluent)

-40 -50 -60

J

--- conventional ....-.Grade Accelsmied Glade

and Oncology

41 (1996)

109-118

also successfully predicts the timing of the onsets of and recoveries from the grade II and 111reactions for patients treated on the low dose rate protocol, as well as the proportions of patients treated at each TDA who reach a grade III mucositis. This would lend credence to the suggestion that the regenerative response is linked quantitatively to the degree of mucosal cellular depletion regardless of the rate at which the depletion occurs. However, it fails to predict the timing of the recovery of reactions caused by the accelerated regime. This is possibly because the mucosal cell kill produced by this regime is clearly sufficient to cause complete mucosal cellular denudation in some cases. In these patients, mucosal recovery occurs as a result of regeneration inwards from the field edges causing the healing process to be prolonged. Some suggestion of this is seen in the time course depictions presented in Fig. 4. The true value of the model, in our opinions, is to serve as a starting point for investigation of the processes involved. These, we suspect, will show that the processes are complex and are not amenable to modelling using parameters simple enough to use in the clinic. Many readers might wonder how a regenerative response capable of making good damage in the range 1.5-2 Gylday can be possible from a mucosa in which the germinal cell population is massively depleted by radiation. Dorr, Kummermehr and colleagues have postulated that the normal restriction to asymmetric division of cells in the mucosal stem cell compartment is released after a certain amount of radiation damage is appreciated by the tissue [4]. A number of ‘abortive’ symmetrical divisions then occurs before these doomed cells exfoliate, but the net effect is a massive increase in the proliferative compartment and maintenance of some mucosal cover over the submucosa (see Fig. 6). This hypothesis seems entirely plausible to us, but is obviously one that can not be modelled easily. Two further points to emerge from our data deserve further discussion. The first is the need for some refinement of the accepted teaching that: ‘Rapid renewal systems are organized in a hierarchical way (H-type), with 3 compartments: stem cells, the

II III

Fig. 5. A model of the reaction time courses of orophatyngeal reactions caused by the ‘conventional’ and ‘accelerated’ TROG trial protocols using Mater data as a basis. The assumptions made are that regeneration is an exponential process which commences when a specific level of cellular depletion is appreciated by the mucosa and which proceeds at a specific rate regardless of the rate at which dose is accumulated. A good fit to the Mater onset data is provided if regeneration commences once 32 ERD units of damage has accumulated and then proceeds at 1.56 ERD units/day (i.e. 1.5-1.8 Gylday) until 38 units are accumulated. At this point, an increase in the rate of regeneration to 1.79 units/day (i.e. 2 Gylday) is depicted. The lines define the evolution of reactions for the ‘median responder’. Half of all patients develop reactions which appear earlier, are charactetised by lower levels of cellular depletion and recover later. The remaining half experience less severe reactions.

Fig. 6. A diagrammatic illustration of the removal of the normal restriction to symmetric stem cell division that occurs when cellular depletion is appreciated by the stem cell compartment according to D&r, Kummermehr and colleagues. The effect is a considerable expansion of the proliferative compartment after the damaged daughter cells have gone through a number of abortive symmetric divisions. The first diagram represents ‘normal (undamaged) steady state’ and the second the situation after the stem cell compartment has started to respond to its depletion.

J.K? Denham

et al. I Radiotherapy

proliferating amplification compartment, and mature functional cells. H-type tissues are characterized by the following factors: (A) Time-to-response is dose-independent and related to the life-span of the functional cells. (B) Rate of recovery is dose-dependent and related to the number of surviving stem cells.’ [17]. [From The ESTRO Book of Basic Clinical Radiobiology.] Our data indicate quite clearly that the timing of the manifestation of acute normal tissue reactions do depend on the rate at which dose is accumulated and that this dependence is in the clinically relevant range of RDAs (e.g. fractionation regimes from 1.8 Gy/day given once daily through to regimes employing 3 Gy/day given once a day or 3.6 Gylday given as two fractions of 1.8 Gy). We suspect that this dependence is up to a point (perhaps 7-8 Gy/day) only. At RDAs that exceed this point, time to expression of damage becomes independent of RDA as the teaching suggests. However, this point is beyond the range of RDAs used in the clinic. The second point is the implication of the quantitative differences in reaction scoring between the Mater Hospital and all other centres for the normal tissue data collection process in multicentre clinical trials. The inter-observer variation responsible for these differences was well shown by the results of a mucosal reaction slide show ‘quiz’ at the eighth annual meeting of TROG (Launceston, September 1995). This quiz suggested that the difference between grade II and III reactions causes participant clinicians the most difficulty, and has lead TROG to consider new strategies to improve the uniformity of reaction reporting.

and Oncology

41 (1996)

109-118

117

damage accumulated at between 1.5 and 2 Gy/day in most cases.

Appendix

1

In the incomplete repair model [14] where complete repair between fractions (exposures) is not assumed, Dale et al. [2] have shown that the biological effect quantity referred to by Barendsen as ‘Extrapolated Response Dose’ or ERD is

where =Total dose = Dose rate = Fraction number = Duration of each exposure ,U = 0.693/t+ (t+ is half time of repair) y = l-epT

D R n T

and S=

nK-K-nK2z+Kni’z”

(1 - Kz)’

where K=eewLX (x is duration of inter-exposure interval) and z=efiT.

Appendix

2

5. Conclusions

The data presented suggest that:

The linear quadratic model without a time correction factor fails to provide an adequate prediction of the frequency or severity of acute mucosal reactions. Time to expression of acute mucosal reaction is subject to substantial inter-patient variation. Acute oropharyngeal mucosal reactions evolve more rapidly than hypopharyngeal reactions. The rate of evolution of acute mucosal reactions depends on the rate of dose accumulation for fractionation schemes commonly used in the clinic. A regenerative mucosal response commences before a patchy grade II mucosal reaction becomes manifest, i.e. as early as the second week of therapy. The timing and magnitude of the regenerative response may be linked to the degree of mucosal cellular depletion regardless of the rate at which depletion occurs. The regenerative response can probably make good

Fowler [7] and others have proposed that a time correction factor for the linear quadratic formula can be derived by assuming exponential regeneration, If T, represents the average doubling time of tissue-rescuing units in days, then daily regeneration can be expressed in ERD units by -A ff where A = Ln2l T, . If T represents the duration of radiotherapy in days and T, represents the number of days after the commencement of treatment that the regenerative process commences, the ERD value of the total treatment course is reduced by: ACT - Td CY

or

Ln2(T - Tk)

“TP If complete repair is assumed, n is fraction number and d fractional dose, then ERD becomes

=,d(l+&)-

IcTiTk’

118

J.U? Denham

et al. I Radiotherapy

In the model we have derived, and depicted in Fig. 5, accumulation of ‘damage’ expressed in cumulative ERD units ‘remembered’ by the tissue is calculated on a daily basis. Until 32 units have accumulated, no regenerative response is assumed and the number of ERD units accumulated from each fraction of treatment is given by

41+$). After 32 units a daily regenerative response (h/o!) of 1.56 ERD units/day and then at 38 units, 1.79 ERD units/day subtracted from the daily accumulated ERD units for both regimes provides a fit for the clinical data. NB. The values of the two parameters defined above are uncertain as is their relationship to the clinically visible endpoints-the reaction grade. The values quoted are illustrative ones derived by iteration and merely indicate that the model can provide a fit to the data.

References [l]

[2]

[3]

[4]

[5]

Barendsen, G.W. Dose fractionation, dose rate and iso-effect relationships for normal tissue responses. Int. .I. Rad. Oncol. Biol. Phys. 8: 1981-1997, 1982. Dale, R.G., Huczkowski, J. and Trott, K. Possible dose rate dependence of recovery kinetics as deduced from a preliminary analysis of the effects of fractionated irradiations at varying dose rates. Br. .I. Radiol. 61: 153-157, 1988. Denham, J.W., Hamilton, C.S., Simpson, S.A., O’Brien, M.Y., Ostwald, P.M., Kron, T. and Dear, K.B.G. Acute reaction parameters for human oropharyngeal mucosa. Radiother. Oncol. 35: 129-137, 1995. Dorr, W., Emmendorfer, H., Haide, E. and Kummermehr, J. Proliferation equivalent of ‘accelerated repopulation’ in mouse oral mucosa. Int. J. Rad. Biol. 66: 157-167, 1994. Dorr, W., Jacubek, A., Kummermehr, J., Herrmann, T., DollingJochem, I. and Eckelt, U. Effects of stimulated repopulation on oral

and Oncology

41 (1996)

109-118

mucositis during conventional radiotherapy. Radiother. Oncol. 37: 100-107, 1995. El Dorr, W. and Weber-Frisch, M. Repopulation response of mouse oral mucosa during unconventional radiotherapy protocols. Radiother. Oncol. 37: 230-236, 1995. [71 Fowler, J.F. What do we need to know to predict the effectiveness of fractionated radiotherapy schedules? In: Prediction of response in radiation therapy: the physical and biological basis, pp. l-13. Editors: B.R. Paliwal, J.F. Fowler, D.E. Herbert, T.J. Kinsella and C.G. Orton. New York, Am. Assoc. Inst. Phys. Inc. 1989. WI Kaanders, J.H.A.M., Pop, L.A.M., Marres, H.A.M., van der Kogel, A.J. and van Daal, W.A.J. Radiotherapy with carbogen breathing and nicotinamide in head and neck cancer: feasibility and toxicity. Radiother. Oncol. 37: 190-198, 1995. Pop, L.A.M., Muytjens, H.L., Bruaset, I. and [91 Kaanders, J.H.A.M., van Daal, W.A.J. Topical administration of antimicrobial agents to prevent irradiation mucositis of the oral cavity and oropharynx: a pilot study. Rad. One. Invest. 2: 283-290, 1995. UOILamb, D.S., Spry, N.A., Gray, A.J., Johnson, A.D., Alexander, S.R., and Dally, M.J. Accelerated fractionation radiotherapy for advanced head and neck cancer. Radiother. Oncol. 18: 107-116, 1990. Ull Liversage, W.E. A critical look at the ret. Br. J. Radiol. 44:91-100, 1971. WI Nyman, J. and Turesson, I. Changes in the basal cell density in pig skin after single radiation doses with different dose rates. Acta Oncol. 30: 753-759, 1991. and mucosal P31 Rugg, T., Saunders, M.I. and Dische, S. Smoking reactions to radiotherapy. Br. J. Radiol. 63: 554-556, 1990. model for survival after u41 Thames, H.D. An ‘incomplete-repair’ fractionated and continuous irradiations. Int. J. Rad. Biol. 47: 319339, 1985. WI Thames, H.D., Bentzen, SM., Turesson, I., Overgaard, M. and Van den Bogaert, W. Time-dose factors in radiotherapy: a review of the human data. Radiother. Oncol. 19: 219-235, 1990. 1161Trott, K. and Kummermehr, J. The time factor and repopulation in tumors and normal tissues. Semin. Rad. Oncol. 3: 115-125, 1993. in normal tissues. In The ESTRO [I71 van der Kogel, A.J. Proliferation Book of Basic Clinical Radiobiology. ESTRO Teaching Course. pp. 21-25. Editor: G.G. Steel. Aarhus, Denmark, 1992. H81 van der Schueren, E., Van den Bogaert, W., Vanuytsel, L. and Van Limbergen, E. Radiotherapy by multiple fractions per day (MFD) in head and neck cancer: acute reactions of skin and mucosa. Int. J. Rad. Oncol. Biol. Phys. 19: 301-311, 1990.