Advertisement

Genetic Variants Predict Optimal Timing of Radiotherapy to Reduce Side-effects in Breast Cancer Patients

Open AccessPublished:October 30, 2018DOI:https://doi.org/10.1016/j.clon.2018.10.001

      Highlights

      • Radiotherapy for breast cancer in the morning causes worse acute and late toxicity.
      • Alleles of two circadian rhythm genes predict worse outcome in the morning.
      • The results will allow genetically determined chronoradiotherapy.

      Abstract

      Aims

      Radiotherapy is an important treatment for many types of cancer, but a minority of patients suffer long-term side-effects of treatment. Multiple lines of evidence suggest a role for circadian rhythm in the development of radiotherapy late side-effects.

      Materials and methods

      We carried out a study to examine the effect of radiotherapy timing in two breast cancer patient cohorts. The retrospective LeND cohort comprised 535 patients scored for late effects using the Late Effects of Normal Tissue-Subjective Objective Management Analytical (LENT-SOMA) scale. Acute effects were assessed prospectively in 343 patients from the REQUITE study using the CTCAE v4 scales. Genotyping was carried out for candidate circadian rhythm variants.

      Results

      In the LeND cohort, patients who had radiotherapy in the morning had a significantly increased incidence of late toxicity in univariate (P = 0.03) and multivariate analysis (P = 0.01). Acute effects in the REQUITE group were also significantly increased in univariate analysis after morning treatment (P = 0.03) but not on multivariate analysis. Increased late effects in the LeND group receiving morning radiotherapy were associated with carriage of the PER3 variable number tandem repeat 4/4 genotype (P = 6 × 10−3) and the NOCT rs131116075 AA genotype (P = 5 × 10−3).

      Conclusion

      Our results suggest that it may be possible to reduce toxicity associated with breast cancer radiotherapy by identifying gene variants that affect circadian rhythm and scheduling for appropriate morning or afternoon radiotherapy.

      Key words

      Introduction

      Circadian rhythms are 24 h endogenous biological cycles that infiltrate every aspect of physiology, biochemistry and behaviour in higher organisms. The phasing of these rhythms can be adjusted by environmental factors such as light and temperature, feeding times and genotoxic agents. Circadian timing regulates rhythmic events in the cell cycle, DNA repair, apoptosis and in the immune system [
      • Innominato P.F.
      • Roche V.P.
      • Palesh O.G.
      • Ulusakarya A.
      • Spiegel D.
      • Levi F.A.
      The circadian timing system in clinical oncology.
      ].
      There is strong evidence that the response to xenobiotics, including drugs, varies according to circadian rhythms, both in terms of efficacy and side-effects [
      • Ballesta A.
      • Innominato P.F.
      • Dallmann R.
      • Rand D.A.
      • Levi F.A.
      Systems chronotherapeutics.
      ]. This has been particularly demonstrated for some anti-cancer agents [
      • Levi F.
      • Okyar A.
      • Dulong S.
      • Innominato P.F.
      • Clairambault J.
      Circadian timing in cancer treatments.
      ,
      • Dulong S.
      • Ballesta A.
      • Okyar A.
      • Levi F.
      Identification of circadian determinants of cancer chronotherapy through in vitro chronopharmacology and mathematical modeling.
      ,
      • Zeng Z.L.
      • Luo H.Y.
      • Yang J.
      • et al.
      Overexpression of the circadian clock gene Bmal1 increases sensitivity to oxaliplatin in colorectal cancer.
      ,
      • Altinok A.
      • Levi F.
      • Goldbeter A.
      Identifying mechanisms of chronotolerance and chronoefficacy for the anticancer drugs 5-fluorouracil and oxaliplatin by computational modeling.
      ].
      The main tissue at risk for acute and late effects of breast radiotherapy is the skin and subcutaneous connective tissue. Fibroblasts in the dermis and keratinocytes in the epidermis show diurnal variation in cell proliferation activity regulated by circadian rhythm genes [
      • Janich P.
      • Meng Q.J.
      • Benitah S.A.
      Circadian control of tissue homeostasis and adult stem cells.
      ]. The skin is subject to damage from the ultraviolet irradiation from the sun. It has been postulated that the pattern of skin cellular proliferation has adapted to this potentially harmful effect during daylight hours and is under circadian control [
      • Beri K.
      • Milgraum S.S.
      Rhyme and reason: the role of circadian rhythms in skin and its implications for physicians.
      ]. Although ultraviolet irradiation only penetrates as far as the dermis it has similar effects on the epidermis as much more penetrating X or gamma rays. Proliferating cells are more radiosensitive or ultraviolet sensitive than non-dividing cells (in G0) and radiosensitivity varies throughout the cell cycle. Cells in the DNA synthetic phase (S phase) are the least sensitive and cells just before mitosis or during mitosis are the most sensitive (G2/M phases) [
      • Hall E.J.
      • Giaccia A.J.
      Radiosensitivity and cell age in the mitotic cycle. Radiobiology for the radiologist.
      ].
      In humans, the maximum number of dividing epidermal cells are in the radioresistant S phase at the time of maximal potential solar exposure [
      • Beri K.
      • Milgraum S.S.
      Rhyme and reason: the role of circadian rhythms in skin and its implications for physicians.
      ]. Janich and colleagues [
      • Janich P.
      • Meng Q.J.
      • Benitah S.A.
      Circadian control of tissue homeostasis and adult stem cells.
      ] identified five peaks of circadian gene activity and proliferation events in human keratinocytes in a 24 h cycle. Three of the observed peaks were in the late evening and early morning hours and were related to keratinocyte differentiation. The other two peaks were in the afternoon and evening hours and were related to DNA replication and cell division. The family of Period genes (PER1–3) were observed to be part of the auto-regulated feedback loop involved in this process.
      There has previously been mixed evidence for whether radiotherapy side-effects are affected by the time of treatment [
      • Chan S.
      • Rowbottom L.
      • McDonald R.
      • Bjarnason G.A.
      • Tsao M.
      • Danjoux C.
      • et al.
      Does the time of radiotherapy affect treatment outcomes? A review of the literature.
      ]. Noh et al. [
      • Noh J.M.
      • Choi D.H.
      • Park H.
      • Huh S.J.
      • Park W.
      • Seol S.W.
      • et al.
      Comparison of acute skin reaction following morning versus late afternoon radiotherapy in patients with breast cancer who have undergone curative surgical resection.
      ] examined the relationship with radiotherapy treatment time and acute skin toxicity in 395 breast cancer patients. They reported that patients treated in the afternoon had more chance of developing acute skin toxicity (Radiation Therapy Oncology Group grade 2 or more) (P = 0.0088). Bjarnason et al. [
      • Bjarnason G.A.
      • Mackenzie R.G.
      • Nabid A.
      • Hodson I.D.
      • El-Sayed S.
      • Grimard L.
      • et al.
      Comparison of toxicity associated with early morning versus late afternoon radiotherapy in patients with head-and-neck cancer: a prospective randomized trial of the National Cancer Institute of Canada Clinical Trials Group (HN3).
      ] investigated the relationship of treatment time to grade 3 or more mucositis in head and neck patients treated with radical doses of radiotherapy but failed to show a significant difference between morning and afternoon groups. However, on subgroup analysis, when patients were divided by gender there was a trend that women had enhanced toxicity in the morning, whereas men showed the effect in the afternoon. A gender-specific circadian rhythm response has also been seen for chemotherapy response [
      • Ahowesso C.
      • Li X.M.
      • Zampera S.
      • Peteri-Brunbäck B.
      • Dulong S.
      • Beau J.
      • et al.
      Sex and dosing-time dependencies in irinotecan-induced circadian disruption.
      ,
      • Giacchetti S.
      • Dugué P.A.
      • Innominato P.F.
      • Bjarnason G.A.
      • Focan C.
      • Garufi C.
      • et al.
      ARTBC International Chronotherapy Group. Sex moderates circadian chemotherapy effects on survival of patients with metastatic colorectal cancer: a meta-analysis.
      ].
      A variable number tandem repeat (VNTR) polymorphism in the PER3 gene has been found to be associated with sleep–wake patterns, with the 5 allele associated with morningness [
      • Archer S.N.
      • Robilliard D.L.
      • Skene D.J.
      • Smits M.
      • Williams A.
      • Arendt J.
      • et al.
      A length polymorphism in the circadian clock gene Per3 is linked to delayed sleep phase syndrome and extreme diurnal preference.
      ]. Earlier evidence found that the C allele of rs1801260 SNP in the CLOCK gene was associated with eveningness [
      • Katzenberg D.
      • Young T.
      • Finn L.
      • Lin L.
      • King D.P.
      • Takahashi J.S.
      • et al.
      A CLOCK polymorphism associated with human diurnal preference.
      ].
      A genome-wide association study for breast cancer side-effects found an association between adverse reactions and a single nucleotide polymorphism (SNP) in a circadian rhythm gene called Nocturnin (NOCT), which encodes a RNA deadenylase [
      • Barnett G.C.
      • Thompson D.
      • Fachal L.
      • Kerns S.
      • Talbot C.
      • Elliott R.M.
      • et al.
      A genome wide association study (GWAS) providing evidence of an association between common genetic variants and late radiotherapy toxicity.
      ]. The SNP was the top hit associated with overall radiosensitivity with a P value of 1.21 × 10−6, which, however, does not meet the conventional significance cut-off for a genome-wide association study. These data led us to investigate potential circadian effects on radiotherapy treatment.

      Materials and Methods

       Patient Selection and Assessment

      All patients were recruited after gaining appropriate consent and ethics approval was granted by the National Health Service research and ethics committee and health research authority.

       LeND Cohort

      For the assessment of late radiotherapy effects, a group of 664 breast cancer patients previously recruited to the LeND study were assessed as previously described [
      • Talbot C.J.
      • Tanteles G.A.
      • Barnett G.C.
      • Burnet N.G.
      • Chang-Claude J.
      • Coles C.E.
      • et al.
      A replicated association between polymorphisms near TNFalpha and risk for adverse reactions to radiotherapy.
      ,
      • Murray R.J.
      • Tanteles G.A.
      • Mills J.
      • Perry A.
      • Peat I.
      • Osman A.
      • et al.
      Association between single nucleotide polymorphisms in the DNA repair gene LIG3 and acute adverse skin reactions following radiotherapy.
      ,
      • Andreassen C.N.
      • Rosenstein B.S.
      • Kerns S.L.
      • Ostrer H.
      • De Ruysscher D.
      • Cesaretti J.A.
      • et al.
      Individual patient data meta-analysis shows a significant association between the ATM rs1801516 SNP and toxicity after radiotherapy in 5456 breast and prostate cancer patients.
      ]. Most patients received 50 Gy of radiation in 25 fractions. Participants were recruited at follow-up oncology clinics from 2008 to 2010 in Leicester, Nottingham and Derby at least 3 years after adjuvant radiotherapy treatment (median follow-up time 62 months). Radiotherapy toxicity was recorded using the Late Effects of Normal Tissue-Subjective Objective Management Analytical (LENT-SOMA) criteria [
      • Pavy J.J.
      • Denekamp J.
      • Letschert J.
      • Littbrand B.
      • Mornex F.
      • Bernier J.
      • et al.
      EORTC Late Effects Working Group
      Late effects toxicity scoring: the SOMA scale.
      ]. At the same visit, volunteers donated a blood sample or buccal cheek swab for DNA extraction. Samples of frozen isolated DNA (1 ng/μl) obtained from whole blood were available for volunteers 150–633.

       REQUITE Cohort

      In total, 343 breast cancer patients were recruited at Leicester Royal Infirmary to the international European Union-funded REQUITE study [
      • West C.
      • Azria D.
      • Chang-Claude J.
      • Davidson S.
      • Lambin P.
      • Rosenstein B.
      • et al.
      The REQUITE project: validating predictive models and biomarkers of radiotherapy toxicity to reduce side-effects and improve quality of life in cancer survivors.
      ] between 2014 and 2016. All breast cancer patients underwent a wide local excision before adjuvant whole breast radiotherapy. Most patients received 40 Gy of radiation in 15 fractions. Co-morbidity data were recorded by the clinical team, with depression being defined as having a clinical diagnosis of depression.
      Patients were assessed by a clinician at baseline and again within the last three fractions of radiotherapy to review any acute reactions using an adapted version of the CTCAE v4 scoring system. At the same time as the baseline assessment, a 10 ml sample of fresh whole blood was collected in an EDTA tube. This was transferred to the CIGMR Biobank (Manchester, UK) who isolated DNA using robotic magnetic bead extraction technology. The isolated DNA (20 ng/μl) was then stored at –80 °C.

       Radiotherapy Timing

      Scheduling of radiotherapy treatment in most cases is a result of department capacity and patient request. At the Leicester Royal Infirmary, radiotherapy records were reviewed to obtain the time of treatment for every fraction received. Patients with more than 66% of their radiotherapy before noon were classified as morning treatment; patients who received more than 66% after noon were classified as afternoon treatment; those falling outside these criteria were classified as a mixed group.
      Radiotherapy had been delivered in Leicester, UK, which is at latitude 52.6°N and has daylight savings time. For the purposes of this analysis, seasons were grouped around the solstices, with the darkest half of the year being 20 September to 20 March and the lightest half of the year being 21 March to 19 September.

       DNA Analysis

      Of the 664 LeND patients, DNA was available for genotyping on 508 patients. In the REQUITE cohort, DNA was available for genotyping in 324 of 343 patients.

       PER3

      In both cohorts, the PER3 VNTR region was amplified by polymerase chain reaction (PCR) and agarose gel electrophoresis was carried out. The two alleles are differentiated by an extra 56 bp repeat in the 5 allele: 4/4 VNTR showed a single band at 639 bp, 4/5 genotype showed two bands at 639 and 685 bp 5/5 VNTR showed a single band at 685 bp.
      In the LeND cohort, PCR for PER3 VNTR was successful in 476 patients; in the remaining 32 samples there was inadequate DNA to produce reliable results. In total, 225 patients were 4/4, 191 patients were 4/5 and the remaining 60 patients 5/5 genotype.
      In the REQUITE cohort, PCR was successful in 309 patients. In total, 140 patients were found to have 4/4, 136 4/5 and the remaining 33 patients were 5/5 genotype. Genotyping results were tested and found to be in Hardy–Weinberg equilibrium.

       NOCT rs13116075

      NOCT rs13116075 was genotyped by Taqman assay. Supplementary Figure S1 shows the plots that resulted from this assay.
      In the LeND cohort, PCR for NOCT rs13116075 was successful in 466 patients; in the remaining 42 samples there was inadequate DNA to produce reliable results. In total, 317 patients were AA, 141 were AG and the remaining eight were GG genotypes. In the REQUITE cohort, genotyping was successful in 323 patients; the remaining one patient sample failed to produce reliable results. Genotyping revealed that 233 patients were AA, 80 were AG and 10 were GG. Genotype frequencies were found to be in Hardy–Weinberg equilibrium.

       CLOCK rs1801260

      CLOCK rs1801260 was genotyped by Taqman assay in just the REQUITE cohort. Supplementary Figure S2 shows the plots that resulted from this assay.

       Statistical Analysis

      For acute toxicity (in REQUITE) breast erythema was taken as a surrogate marker of overall acute toxicity. Any baseline score was deducted from the score assessed within the last three fractions of radiotherapy to give a corrected acute toxicity score.
      Late toxicity (in LeND) was assessed using Bivariate STAT score. Use of the STAT score was described by Barnett et al. [
      • Barnett G.C.
      • Thompson D.
      • Fachal L.
      • Kerns S.
      • Talbot C.
      • Elliott R.M.
      • et al.
      A genome wide association study (GWAS) providing evidence of an association between common genetic variants and late radiotherapy toxicity.
      ] and has been used previously as a dichotomised variable [
      • Talbot C.J.
      • Tanteles G.A.
      • Barnett G.C.
      • Burnet N.G.
      • Chang-Claude J.
      • Coles C.E.
      • et al.
      A replicated association between polymorphisms near TNFalpha and risk for adverse reactions to radiotherapy.
      ]. In brief, a Z score was calculated for each patient and all toxicity end points (i.e. fibrosis, telangiectasia, atrophy, oedema) (Z = [score – mean score]/standard deviation for the whole population). A STAT score can then be calculated as an average of the patient Z scores. The population are then divided into upper quartile and lower three-quarters to form a bivariate STAT score.
      Statistical analysis was carried out using IBM SPSS version 24. Differences in categorical variables were analysed by chi-squared test. Multivariable analysis was carried out using logistic regression. Covariates were selected by bidirectional elimination.

      Results

       Radiotherapy Treatment Time

       The LeND Cohort

      In total, 664 women were enrolled in the LeND cohort and of these radiotherapy treatment time was available for 536 patients. Of those with no data, 75 were recruited in subsites and 53 received radiotherapy before computerised records became available in 1998. Patients were grouped according to treatment time: 185 patients (34.5%) received radiotherapy mainly in the morning, 170 patients (31.7%) received radiotherapy mainly in the afternoon and 181 (33.8%) had a mix of treatment times. Patients received either 45 Gy X-rays in 20 fractions or 50 Gy in 25 fractions. Seventy-six patients received a 9–15 Gy electron boost in three to five fractions. Baseline characteristics for these patients by group are shown in Table 1. Chi-squared testing for categorical variables and ANOVA testing for continuous variables revealed that none of these baseline characteristics was significantly different between the treatment time groups. The only exception was if the mixed group were excluded from the analysis, then significantly more patients received a boost in the morning compared with the afternoon (P = 0.04).
      Table 1Baseline tumour, patient and treatment characteristics for the LeND cohort. Percentages are of non-missing data within each time group
      PatientsRadiotherapy treatment timeP =
      AMMixedPMTotal
      185181170536
      Age (mean years)59.157.458.70.28
      Grade129 (16.9%)40 (25.8%)39 (24.8%)108 (22.3%)
      281 (47.1%)73 (47.1%)67 (42.7%)221 (45.7%)0.20
      362 (36.0%)41 (26.5%)51 (32.5%)154 (31.8%)
      Missing13261352
      ChemotherapyYes71 (38.4%)49 (27.1%)49 (29.0%)169 (31.6%)
      No114 (61.6%)132 (72.9%)120 (71.0%)366 (68.4%)0.05
      Missing0011
      Oestrogen receptor positiveYes162 (88.5%)147 (82.1%)141 (82.9%)450 (84.6%)
      No21 (11.5%)32 (17.9%)29 (17.1%)82 (15.4%)0.19
      Missing2103
      Mean radiotherapy dose (Gy)47.847.8480.76
      Boost to tumour bedYes36 (19.6%)22 (12.2%)20 (11.8%)78 (14.6%)
      No148 (80.4%)159 (87.8%)150 (88.2%)457 (85.4%)0.06
      Missing1001
      DiabetesYes15 (8.1%)6 (3.3%)12 (7.1%)33 (6.2%)
      No170 (91.9%)175 (96.7%)158 (92.9%)503 (93.8%)0.14
      SmokerCurrent smoker27 (14.9%)21 (11.7%)16 (9.5%)64 (12.1%)
      No134 (74.0%)138 (76.7%)121 (72.0%)393 (74.3%)0.16
      Ex-smoker20 (11.0%)21 (11.7%)31 (18.5%)72 (13.6%)
      Missing4026
      Side of treatmentRight100 (54.1%)102 (56.4%)90 (52.9%)292 (54.5%)
      Left79 (42.7%)73 (40.3%)78 (45.9%)230 (42.9%)0.59
      Bilateral6 (3.2%)6 (3.3%)2 (1.2%)14 (2.6%)
      Bra cup sizeA–AA13 (9.2%)9 (6.8%)16 (11.9%)38 (9.3%)
      B53 (37.3%)40 (30.0%)41 (30.4%)134 (33.7%)
      C34 (23.9%)36 (27.1%)27 (20.0%)96 (23.6%)0.38
      D22 (15.5%)16 (12.1%)22 (16.3%)60 (14.6%)
      DD–GG20 (14.0%)32 (24.1%)29 (21.4%)71 (19.7%)
      Missing434835126
      Late toxicity (bivariate STAT score) was available on 536 of the patients. Univariate analysis using a chi-squared test was carried out and showed a significantly increased frequency of toxicity in patients treated in the mornings: 29.2% of the morning treatment group had high toxicity compared with 21.1% of the afternoon group and 17.7% of the mixed group (P = 0.03). Multivariable analysis was carried out using logistic regression and radiotherapy treatment time remained significant (P = 0.01). Results are summarised in Table 2. An interaction term for boost × treatment time was included due to the difference between the groups in the proportion receiving boosts.
      Table 2Multivariate analysis (LeND cohort) for effect on late toxicity (bivariate STAT score)
      VariableP ValueOdds ratio95% confidence interval for odds ratio
      LowerUpper
      Boost to tumour bed0.582.320.1245.52
      BED0.0370.541.643044
      Cup size5.1 × 10−71.271.161.39
      Afternoon radiotherapy treatment0.010.610.410.90
      Boost × time1.0 × 10−36.802.2320.63
      BED, biological equivalent dose.

       REQUITE Cohort Baseline Tumour, Patient and Treatment Data

      Radiotherapy treatment time was available for 343 of the REQUITE Leicester breast cohort. In total, 111 patients (32.4%) received radiotherapy in the morning, 152 were treated in the afternoon (44.3%) and 80 (23.3%) received a mix of morning and afternoon treatment. Table 3 summarises the baseline characteristics of the tumour, patient and treatment between the treatment times. Chi-squared testing for categorical variables and ANOVA testing for continuous variables revealed that significantly more patients in the morning group received a boost (P = 0.01) and had higher grade tumours (P = 0.01) compared with the afternoon and mixed groups.
      Table 3Baseline characteristics in the REQUITE cohort for tumour, patient and treatment. Percentages are of non-missing data within each time group
      PatientsRadiotherapy treatment timeP =
      AMMixedPMTotal
      11180152343
      Age (mean years)60.061.460.50.60
      Grade125 (28%)9 (14%)35 (30%)69 (25%)
      236 (40%)38 (58%)65 (55%)139 (51%)
      328 (31%)18 (28%)18 (15%)64 (24%)0.008
      Missing22153471
      ChemotherapyYes4 (4%)6 (9%)11 (9%)21 (7%)
      No87 (96%)62 (91%)111 (91%)260 (93%)0.40
      Missing20123062
      Oestrogen receptor positiveYes50 (55%)35 (51%)68 (55%)150 (54%)
      No41 (45%)33 (49%)56 (45%)130 (46%)0.89
      Missing20123163
      Mean radiotherapy dose (Gy)41.240.841.00.77
      Boost to tumour bedYes12 (13%)5 (7%)3 (2%)20 (8%)
      No79 (87%)63 (93%)119 (98%)261 (93%)0.01
      Missing20123062
      DiabetesYes8 (7%)7 (9%)14 (9%)29 (8%)
      No103 (93%)72 (91%)138 (91%)313 (92%)0.84
      Missing0101
      SmokerCurrent smoker14 (14%)4 (7%)19 (17%)37 (14%)
      No39 (40%)27 (44%)45 (39%)111 (41%)
      Ex-smoker44 (45%)30 (49%)50 (44%)124 (46%)0.47
      Missing14193871
      Side of treatmentRight48 (54%)36 (54%)61 (50%)145 (52%)
      Left41 (46%)31 (46%)61 (50%)133 (48%)0.82
      Bilateral0000
      Missing22133065
      Body mass index29.127.929.70.59
      Bra cup sizeAA–A9 (8.1%)6 (7.6%)8 (5.3%)23 (6.7%)
      B33 (29.7%)19 (24.1%)43 (28.5%)95 (27.9%)
      C24 (21.6%)13 (16.5%)41 (27.2%)78 (22.9%)
      D18 (16.2%)15 (19.0%)27 (17.9%)60 (17.6%)
      DD–J27 (24.3%)26 (32.9%)32 (21.2%)85 (24.9%)0.35
      Missing112
      Acute breast erythema score was available on 331 of the study participants. Univariate analysis using a chi-squared test was carried out and showed a significantly increased toxicity rate in patients treated in the mornings: 23.6% of patients had grade 2 or more erythema in the morning group compared with 11.0% in the afternoon group and 19.0% in the mixed group (P = 0.03). Removal of the mixed treatment time group from the analysis resulted in a P value of 7.0 × 10−3. Multivariable analysis was carried out using logistic regression and radiotherapy treatment time was not significant (P = 0.28). Results are summarised in Table 4.
      Table 4Multivariate analysis of the REQUITE breast cohort for the effect on acute toxicity
      VariableP ValueOdds ratio95% confidence interval for odds ratio
      LowerUpper
      Bra cup size4.90 × 10−51.711.322.23
      Depression0.020.070.010.62
      Boost to tumour bed0.641.660.2014.17
      Boost × radiotherapy time0.491.890.3111.64
      Season0.651.210.532.72
      Afternoon radiotherapy treatment0.280.760.461.25
      Biological equivalent dose1.57 × 10−71.451.261.66

       Genetic Variants as Effect Modifier

      We tested if the time-of-day effect is modified by candidate polymorphisms in circadian rhythm genes, including SNPs in the CLOCK and NOCT genes and a VNTR in the PER3 gene.

       LeND Cohort

      There was no significant direct relationship between PER3 VNTR and late toxicity (bivariate STAT score). Taking the time of radiotherapy into consideration showed a significant effect of PER3 VNTR on late toxicity, with 4/4 PER3 VNTR being associated with increased toxicity if treated in the morning compared with the afternoon (P = 6.0 × 10−3) (Table 5). A Bonferroni correction for six tests (based on six genotypes in Table 5) would give a P value of 0.036.
      Table 5Late toxicity split by PER, NOCT rs13116075 and radiotherapy treatment time in the LeND cohort
      Radiotherapy treatment timePER3 VNTRLate toxicity (bivariate STAT score) n =P Value
      Chi-squared test for significance including all treatment times.
      NOCT rs13116075Late toxicity (bivariate STAT score n =)P Value
      Chi-squared test for significance including all treatment times.
      75% lowest25% highest75% lowest25% highest
      Morning4/443 (68.3%)20 (31.7%)6.0 × 10−3AA61 (76.7%)29 (23.3%)5.0 × 10−3
      Afternoon49 (79.0%)13 (21.0%)80 (85.1%)14 (14.9%)
      Mixed54 (91.5%)5 (8.5%)72 (84.7%)13 (15.3%)
      Morning4/537 (77.1%)11 (22.9%)0.49AG22 (75.9%)7 (24.1%)0.74
      Afternoon51 (85.0%)9 (15.0%)34 (75.6%)11 (24.4%)
      Mixed45 (77.6%)13 (22.4%)36 (81.8%)8 (18.2%)
      Morning5/511 (73.3%)4 (26.7%)0.75GG3 (100%)00.08
      Afternoon12 (66.7%)6 (33.3%)1 (33.3%)2 (66.7%)
      Mixed14 (77.8%)4 (22.2%)00
      Total3168531685
      VNTR, variable number tandem repeat.
      Chi-squared test for significance including all treatment times.
      There was no significant direct relationship with NOC rs13116075 and late toxicity (bivariate STAT score). Taking the time of radiotherapy treatment into consideration showed a significant effect of NOC rs13116075 on late toxicity. Patients carrying the AA NOC rs13116075 genotype had increased toxicity if treated in the mornings (P = 5.0 × 10−3) (Table 5). A Bonferroni correction for six tests would give a P value of 0.03.
      Combination of PER3 VNTR and NOCT rs13116075 genotypes had no significant direct effect on late radiotherapy toxicity. However, taking the time of radiotherapy treatment into consideration showed a significant effect of NOCT rs13116075 and PER3 VNTR on late toxicity. Patients with AA NOCT rs13116075 genotype and 4/4 PER3 VNTR had increased late radiotherapy toxicity if treated in the mornings (P = 4.5 × 10−4) (data not shown).

       REQUITE Cohort

      There was no association between PER3 VNTR genotype and acute toxicity score in the 294 patients with data available for both (P = 0.79). Taking the time of radiotherapy treatment into consideration showed no significant effect of PER3 VNTR on acute toxicity with any of the genotypes (Table 6). If the mixed treatment time group was excluded from the analysis, patients with 4/5 PER3 VNTR had increased acute toxicity with morning compared with afternoon treatment (P = 0.02).
      Table 6Acute breast erythema split by PER3 and NOCT rs13116075 genotyping and radiotherapy treatment time in the REQUITE breast cohort
      Treatment timePER3 VNTRAcute toxicity (breast erythema score) n =P ValueNOCT rs13116075Acute toxicity (breast erythema score) n =P Value
      <2≥2<2≥2
      Morning4/431 (81.6%)7 (18.4%)0.71
      Chi-squared test for significance including all treatment times.
      /0.71
      Chi-squared test for significance excluding the mixed treatment times group.
      AA59 (80.8%)14 (19.2%)0.21
      Chi-squared test for significance including all treatment times.
      /0.18
      Chi-squared test for significance excluding the mixed treatment times group.
      Afternoon49 (84.5%)9 (15.5%)83 (88.3%)11 (11.7%)
      Mixed31 (88.6%)4 (11.4%)49 (90.7%)5 (9.3%)
      Morning4/532 (72.7%)12 (27.3%)0.07
      Chi-squared test for significance including all treatment times.
      /0.02
      Chi-squared test for significance excluding the mixed treatment times group.
      AG15 (65.2%)8 (34.8%)0.06
      Chi-squared test for significance including all treatment times.
      /0.04
      Chi-squared test for significance excluding the mixed treatment times group.
      Afternoon53 (89.8%)6 (10.2%)30 (88.2%)4 (11.8%)
      Mixed20 (76.9%)6 (23.1%)10 (62.5%)6 (37.5%)
      Morning5/59 (81.8%)2 (18.2%)0.19
      Chi-squared test for significance including all treatment times.
      /0.16
      Chi-squared test for significance excluding the mixed treatment times group.
      GG1 (100%)0N/A
      Afternoon10 (100%)05 (100%)0
      Mixed7 (70.0%)3 (30.0%)1 (100%)0
      TOTAL2424925351
      VNTR, variable number tandem repeat.
      Chi-squared test for significance including all treatment times.
      Chi-squared test for significance excluding the mixed treatment times group.
      Acute toxicity score was available for 308 of the 323 breast patients, with genotyping data available for NOC rs13116075. The NOC rs13116075 AG genotype was significantly associated with increased toxicity (P = 0.04), but not after applying a correction for multiple testing. The other genotypes were not related to toxicity.
      Taking the time of radiotherapy treatment into consideration showed no significant effect of NOC rs13116075 (Table 6) on acute toxicity with any of the genotypes. If the mixed treatment time group was excluded from the analysis, patients with AG NOC rs13116075 had increased acute toxicity in the mornings (P = 0.04). A Bonferroni correction for eight tests would render these results non-significant.
      Combination of PER3 VNTR and NOCT rs13116075 had no direct association with increased late toxicity. Nor was there any evidence of the combination modifying the time-of-day effect on toxicity score. There was no effect of CLOCK genotype in any analysis.

      Discussion

      In this study we found some evidence that breast cancer patients treated in the morning had worse radiotherapy side-effects than those treated in the afternoon.
      These results are in contrast with some earlier studies [
      • Chan S.
      • Rowbottom L.
      • McDonald R.
      • Bjarnason G.A.
      • Tsao M.
      • Danjoux C.
      • et al.
      Does the time of radiotherapy affect treatment outcomes? A review of the literature.
      ,
      • Noh J.M.
      • Choi D.H.
      • Park H.
      • Huh S.J.
      • Park W.
      • Seol S.W.
      • et al.
      Comparison of acute skin reaction following morning versus late afternoon radiotherapy in patients with breast cancer who have undergone curative surgical resection.
      ], but the differences are probably due to differences in the gender of patients, irradiated tissue type, geographical latitude, allocation of time groups and methods of booking patients into radiotherapy.
      To enable the reduction of side-effects while maintaining activity in radiotherapy suites over the whole day it is necessary to be able to predict which patients would benefit from having their treatment delivered at a defined time of day (chronotherapy). To that end we carried out genotyping for some candidate circadian rhythm polymorphisms and found that the circadian effect was strongest in individuals who were homozygous for the PER3 4 repeat or NOCT A alleles. Importantly, these alleles were not found to be directly associated with radiotherapy toxicity in this study, but only to potentiate the time-of-day effect. It is possible that larger studies would find a direct association, as the NOCT SNP was previously found to be associated with overall toxicity by genome-wide association study. The PER group of genes are intimately connected with the control and timing of keratinocyte division [
      • Janich P.
      • Meng Q.J.
      • Benitah S.A.
      Circadian control of tissue homeostasis and adult stem cells.
      ]. Nocturnin has been shown to affect the proliferation of adipocytes, an important cell type in the breast [
      • Kawai M.
      • Green C.B.
      • Lecka-Czernik B.
      • Douris N.
      • Gilbert M.R.
      • Kojima S.
      • et al.
      A circadian-regulated gene, Nocturnin, promotes adipogenesis by stimulating PPAR-gamma nuclear translocation.
      ].
      There are several potential physiological mechanisms to explain how time could affect reactions to irradiation, e.g. melatonin, cortisol, inflammatory factors or cell proliferation/DNA damage. Melatonin has antioxidant properties [
      • Ben-David M.A.
      • Elkayam R.
      • Gelernter I.
      • Pfeffer R.M.
      Melatonin for prevention of breast radiation dermatitis: a phase II, prospective, double-blind randomized trial.
      ], has been shown to be radioprotective in mice [
      • Vijayalaxmi
      • Meltz M.L.
      • Reiter R.J.
      • Herman T.S.
      • Kumar K.S.
      Melatonin and protection from whole-body irradiation: survival studies in mice.
      ] and reduce oral mucositis in irradiated rats [
      • Abdel Moneim A.E.
      • Guerra-Librero A.
      • Florido J.
      • Shen Y.Q.
      • Fernandez-Gil B.
      • Acuna-Castroviejo D.
      • et al.
      Oral mucositis: melatonin gel an effective new treatment.
      ]. Cortisol levels can be used as a marker of stress and can affect inflammatory markers [
      • Wright Jr., K.P.
      • Drake A.L.
      • Frey D.J.
      • Fleshner M.
      • Desouza C.A.
      • Gronfier C.
      • et al.
      Influence of sleep deprivation and circadian misalignment on cortisol, inflammatory markers, and cytokine balance.
      ]. Cortisol levels can influence the rate of cell division and may be associated with possible increased cellular division of skin in the morning compared with later in the day [
      • Sporl F.
      • Korge S.
      • Jurchott K.
      • Wunderskirchner M.
      • Schellenberg K.
      • Heins S.
      • et al.
      Kruppel-like factor 9 is a circadian transcription factor in human epidermis that controls proliferation of keratinocytes.
      ].
      Many clinical studies have shown a weak association between acute and late toxicity, with some making a distinction between consequential and non-consequential late effects [
      • Dörr W.
      • Hendry J.H.
      Consequential late effects in normal tissues.
      ]. Therefore, the fact that the time-of-day effect is the same in the present study for acute and late toxicity may give a clue to the mechanism. Future work will determine how the cell cycle of keratinocytes and fibroblasts responds to circadian rhythms.
      There were several limitations of this study. First, the study was conducted at a single centre and therefore the generalisability of the results will not be known until replicated at other centres. Second, acute and late toxicity were assessed in different cohorts who received radiotherapy in different calendar years with different radiotherapy protocols, which complicates interpretation and adjustment for multiple testing. Added to this is that the LeND cohort was collected retrospectively, meaning no assessment was possible of tissue changes from baseline. This will be remedied when late toxicity data become available for the REQUITE cohort. Acute toxicity in the REQUITE cohort was recorded before the end of radiotherapy, so will have missed manifestations occurring some weeks later. Third, the cohort sizes limit the statistical power to detect effects; larger cohorts are needed to confirm the data.
      These genetic data potentially open the possibility of a simple test to identify the patients who would benefit from receiving their treatment in the afternoon, whereas the remaining patients could be treated at any time. This approach will need to be verified using a clinical trial that randomises between a group in which patients choose their own treatment time and a chronotherapy group. This virtually cost-free intervention would be predicted to reduce side-effects and improve quality of life for breast cancer survivors.

      Conflicts of interest

      The authors declare no conflict of interest.

      Acknowledgments

      We would like to thank the patients in both the LeND and the REQUITE studies for their involvement in this work, as well as Anusha Müller and Rebecca Elliott for excellent data and project management of the REQUITE project. Support for this study was provided by Breast Cancer Now (grant number 2007NovPR45 ), Hope Against Cancer (grant number RM33G0351 ) and the European Union (grant agreement 601826 ). Patient recruitment was supported through the Leicester Experimental Cancer Centre . Taqman genotyping was carried by the NUCLEUS Genomics service at the University of Leicester. The REQUITE consortium includes: David Azria, Anthony Brookes, Tom Burr, Jenny Chang-Claude, Susan Davidson, Dirk De Ruysscher, Alison Dunning, Rebecca Elliott, Sara Gutiérrez Enríquez, Philippe Lambin, Tiziana Rancati, Barry Rosenstein, Petra Seibold, R. Paul Symonds, Chris Talbot, Hubert Thierens, Riccardo Valdagni, Ana Vega, Liv Veldeman, Frederik Wenz, Martin Yuille and Catharine West.

      Appendix A. Supplementary data

      Fig S1
      Fig S2

      References

        • Innominato P.F.
        • Roche V.P.
        • Palesh O.G.
        • Ulusakarya A.
        • Spiegel D.
        • Levi F.A.
        The circadian timing system in clinical oncology.
        Ann Med. 2014; 46: 191-207
        • Ballesta A.
        • Innominato P.F.
        • Dallmann R.
        • Rand D.A.
        • Levi F.A.
        Systems chronotherapeutics.
        Pharmacol Rev. 2017; 69: 161-199
        • Levi F.
        • Okyar A.
        • Dulong S.
        • Innominato P.F.
        • Clairambault J.
        Circadian timing in cancer treatments.
        Annu Rev Pharmacol Toxicol. 2010; 50: 377-421
        • Dulong S.
        • Ballesta A.
        • Okyar A.
        • Levi F.
        Identification of circadian determinants of cancer chronotherapy through in vitro chronopharmacology and mathematical modeling.
        Mol Cancer Ther. 2015; 14: 2154-2164
        • Zeng Z.L.
        • Luo H.Y.
        • Yang J.
        • et al.
        Overexpression of the circadian clock gene Bmal1 increases sensitivity to oxaliplatin in colorectal cancer.
        Clin Cancer Res. 2014; 20: 1042-1052
        • Altinok A.
        • Levi F.
        • Goldbeter A.
        Identifying mechanisms of chronotolerance and chronoefficacy for the anticancer drugs 5-fluorouracil and oxaliplatin by computational modeling.
        Eur J Pharm Sci. 2009; 36: 20-38
        • Janich P.
        • Meng Q.J.
        • Benitah S.A.
        Circadian control of tissue homeostasis and adult stem cells.
        Curr Opin Cell Biol. 2014; 31: 8-15
        • Beri K.
        • Milgraum S.S.
        Rhyme and reason: the role of circadian rhythms in skin and its implications for physicians.
        Future Sci OA. 2016; 2: Fso115
        • Hall E.J.
        • Giaccia A.J.
        Radiosensitivity and cell age in the mitotic cycle. Radiobiology for the radiologist.
        7 ed. Lippincott Williams and Wilkins, Philadelphia2012: 54-66
        • Chan S.
        • Rowbottom L.
        • McDonald R.
        • Bjarnason G.A.
        • Tsao M.
        • Danjoux C.
        • et al.
        Does the time of radiotherapy affect treatment outcomes? A review of the literature.
        Clin Oncol. 2017; 29: 231-238
        • Noh J.M.
        • Choi D.H.
        • Park H.
        • Huh S.J.
        • Park W.
        • Seol S.W.
        • et al.
        Comparison of acute skin reaction following morning versus late afternoon radiotherapy in patients with breast cancer who have undergone curative surgical resection.
        J Radiat Res. 2014; 55: 553-558
        • Bjarnason G.A.
        • Mackenzie R.G.
        • Nabid A.
        • Hodson I.D.
        • El-Sayed S.
        • Grimard L.
        • et al.
        Comparison of toxicity associated with early morning versus late afternoon radiotherapy in patients with head-and-neck cancer: a prospective randomized trial of the National Cancer Institute of Canada Clinical Trials Group (HN3).
        Int J Radiat Oncol Biol Phys. 2009; 73: 166-172
        • Ahowesso C.
        • Li X.M.
        • Zampera S.
        • Peteri-Brunbäck B.
        • Dulong S.
        • Beau J.
        • et al.
        Sex and dosing-time dependencies in irinotecan-induced circadian disruption.
        Chronobiol Int. 2011; 28: 458-470
        • Giacchetti S.
        • Dugué P.A.
        • Innominato P.F.
        • Bjarnason G.A.
        • Focan C.
        • Garufi C.
        • et al.
        ARTBC International Chronotherapy Group. Sex moderates circadian chemotherapy effects on survival of patients with metastatic colorectal cancer: a meta-analysis.
        Ann Oncol. 2012; 23: 3110-3116
        • Archer S.N.
        • Robilliard D.L.
        • Skene D.J.
        • Smits M.
        • Williams A.
        • Arendt J.
        • et al.
        A length polymorphism in the circadian clock gene Per3 is linked to delayed sleep phase syndrome and extreme diurnal preference.
        Sleep. 2003; 26: 413-415
        • Katzenberg D.
        • Young T.
        • Finn L.
        • Lin L.
        • King D.P.
        • Takahashi J.S.
        • et al.
        A CLOCK polymorphism associated with human diurnal preference.
        Sleep. 1998; 21: 569-576
        • Barnett G.C.
        • Thompson D.
        • Fachal L.
        • Kerns S.
        • Talbot C.
        • Elliott R.M.
        • et al.
        A genome wide association study (GWAS) providing evidence of an association between common genetic variants and late radiotherapy toxicity.
        Radiother Oncol. 2014; 111: 178-185
        • Talbot C.J.
        • Tanteles G.A.
        • Barnett G.C.
        • Burnet N.G.
        • Chang-Claude J.
        • Coles C.E.
        • et al.
        A replicated association between polymorphisms near TNFalpha and risk for adverse reactions to radiotherapy.
        Br J Cancer. 2012; 107: 748-753
        • Murray R.J.
        • Tanteles G.A.
        • Mills J.
        • Perry A.
        • Peat I.
        • Osman A.
        • et al.
        Association between single nucleotide polymorphisms in the DNA repair gene LIG3 and acute adverse skin reactions following radiotherapy.
        Radiother Oncol. 2011; 99: 231-234
        • Andreassen C.N.
        • Rosenstein B.S.
        • Kerns S.L.
        • Ostrer H.
        • De Ruysscher D.
        • Cesaretti J.A.
        • et al.
        Individual patient data meta-analysis shows a significant association between the ATM rs1801516 SNP and toxicity after radiotherapy in 5456 breast and prostate cancer patients.
        Radiother Oncol. 2016; 121: 431-439
        • Pavy J.J.
        • Denekamp J.
        • Letschert J.
        • Littbrand B.
        • Mornex F.
        • Bernier J.
        • et al.
        • EORTC Late Effects Working Group
        Late effects toxicity scoring: the SOMA scale.
        Radiother Oncol. 1995; 35: 11-15
        • West C.
        • Azria D.
        • Chang-Claude J.
        • Davidson S.
        • Lambin P.
        • Rosenstein B.
        • et al.
        The REQUITE project: validating predictive models and biomarkers of radiotherapy toxicity to reduce side-effects and improve quality of life in cancer survivors.
        Clin Oncol. 2014; 26: 739-742
        • Kawai M.
        • Green C.B.
        • Lecka-Czernik B.
        • Douris N.
        • Gilbert M.R.
        • Kojima S.
        • et al.
        A circadian-regulated gene, Nocturnin, promotes adipogenesis by stimulating PPAR-gamma nuclear translocation.
        Proc Natl Acad Sci USA. 2010; 107: 10508-10513
        • Ben-David M.A.
        • Elkayam R.
        • Gelernter I.
        • Pfeffer R.M.
        Melatonin for prevention of breast radiation dermatitis: a phase II, prospective, double-blind randomized trial.
        Isr Med Assoc J. 2016; 18: 188-192
        • Vijayalaxmi
        • Meltz M.L.
        • Reiter R.J.
        • Herman T.S.
        • Kumar K.S.
        Melatonin and protection from whole-body irradiation: survival studies in mice.
        Mutat Res. 1999; 425: 21-27
        • Abdel Moneim A.E.
        • Guerra-Librero A.
        • Florido J.
        • Shen Y.Q.
        • Fernandez-Gil B.
        • Acuna-Castroviejo D.
        • et al.
        Oral mucositis: melatonin gel an effective new treatment.
        Int J Mol Sci. 2017; 18: 1003
        • Wright Jr., K.P.
        • Drake A.L.
        • Frey D.J.
        • Fleshner M.
        • Desouza C.A.
        • Gronfier C.
        • et al.
        Influence of sleep deprivation and circadian misalignment on cortisol, inflammatory markers, and cytokine balance.
        Brain Behav Immun. 2015; 47: 24-34
        • Sporl F.
        • Korge S.
        • Jurchott K.
        • Wunderskirchner M.
        • Schellenberg K.
        • Heins S.
        • et al.
        Kruppel-like factor 9 is a circadian transcription factor in human epidermis that controls proliferation of keratinocytes.
        Proc Natl Acad Sci USA. 2012; 109: 10903-10908
        • Dörr W.
        • Hendry J.H.
        Consequential late effects in normal tissues.
        Radiother Oncol. 2001; 61: 223-231