Exposure of the heart in lung cancer radiation therapy: A systematic review of heart doses published during 2013 to 2020

Background and purpose: Lung cancer radiotherapy increases the risk of cardiotoxicity and heart radiation dose is an independent predictor of poor survival. This study describes heart doses and strategies aiming to reduce exposure. Materials and methods: A systematic review of lung cancer dosimetry studies reporting heart doses published 2013–2020 was undertaken. Doses were compared according to laterality, region irradiated, treat- ment modality (stereotactic ablative body radiotherapy (SABR) and non-SABR), planning technique, and respiratory motion management. Results: For 392 non-SABR regimens in 105 studies, the average MHD was 10.3 Gy (0.0–48.4) and was not signiﬁcantly different between left and right-sided tumours. It was similar between IMRT and 3DCRT (10.9 Gy versus 10.6 Gy) and lower with particle beam therapy (proton 7.0 Gy; carbon-ion 1.9 Gy). Active respiratory motion management reduced exposure (7.4 Gy versus 9.3 Gy). For 168 SABR regimens in 35 studies, MHD was 4.0 Gy (0.0–32.4). Exposure was higher in central and lower lobe lesions (6.3 and 5.8 Gy respectively). MHD was lowest for carbon ions (0.5 Gy) compared to other techniques. Active res- piratory motion management reduced exposure (2.4 Gy versus 5.0 Gy). Delineation guidelines and Dose Volume Constraints for the heart varied substantially. Conclusions: There is scope to reduce heart radiation dose in lung cancer radiotherapy. Consensus on planning objectives, contouring and DVCs for the heart may lead to reduced heart doses in the future. For IMRT, more stringent optimisation objectives may reduce heart dose. Active respiratory motion management or particle therapy may be considered in situations where cardiac dose is high.


a b s t r a c t
Background and purpose: Lung cancer radiotherapy increases the risk of cardiotoxicity and heart radiation dose is an independent predictor of poor survival. This study describes heart doses and strategies aiming to reduce exposure. Materials and methods: A systematic review of lung cancer dosimetry studies reporting heart doses published 2013-2020 was undertaken. Doses were compared according to laterality, region irradiated, treatment modality (stereotactic ablative body radiotherapy (SABR) and non-SABR), planning technique, and respiratory motion management. Results: For 392 non-SABR regimens in 105 studies, the average MHD was 10.3 Gy (0.0-48.4) and was not significantly different between left and right-sided tumours. It was similar between IMRT and 3DCRT (10.9 Gy versus 10.6 Gy) and lower with particle beam therapy (proton 7.0 Gy; carbon-ion 1.9 Gy). Active respiratory motion management reduced exposure (7.4 Gy versus 9.3 Gy). For 168 SABR regimens in 35 studies, MHD was 4.0 Gy (0.0-32.4). Exposure was higher in central and lower lobe lesions (6.3 and 5.8 Gy respectively). MHD was lowest for carbon ions (0.5 Gy) compared to other techniques. Active respiratory motion management reduced exposure (2.4 Gy versus 5.0 Gy). Delineation guidelines and Dose Volume Constraints for the heart varied substantially. Conclusions: There is scope to reduce heart radiation dose in lung cancer radiotherapy. Consensus on planning objectives, contouring and DVCs for the heart may lead to reduced heart doses in the future. For IMRT, more stringent optimisation objectives may reduce heart dose. Active respiratory motion management or particle therapy may be considered in situations where cardiac dose is high. Lung cancer is the most common cancer worldwide with an incidence of 2.2 million reported in 2020 [1]. Radical radiotherapy (RT) is used to treat locally advanced non-small cell lung cancer (NSCLC), early-stage NSCLC not suitable for surgery, and limited stage small cell lung cancer (SCLC). Incidental exposure of the heart is unavoidable in most patients, and this may increase the risk of cardiac disease including ischaemic heart disease, heart failure, valvular heart disease, pericardial disease, and conduction system abnormalities [2,3]. Injury to the heart can manifest clinically many years after RT [4] but acute cardiotoxicity may also occur within a few months post treatment, particularly in patients with lung cancer [5,6]. The association between cardiac radiation exposures and subsequent cardiotoxicity is well established for patients undergoing RT for breast cancer [7,8] and Hodgkin's lymphoma [9][10][11] and cardiac-sparing techniques are routinely incorporated into treatment planning and delivery [12,13]. Given the poor prognosis traditionally associated with lung cancer, most toxicity studies have focused on acute radiation-related pneumonitis [14][15][16] and oesophagitis [17][18][19]. However, cardiac toxicity is a real concern for this group of patients. Lung cancer survival is improving due to early detection, tobacco control interventions and improved treatments. The cohort of patients involved are typically greater than 60 years and may have a history of cardiac disease or preexisting factors for developing cardiac disease e.g., smoking, hypertension [20,21]. There is increasing evidence available on the risk of cardiac morbidity and mortality relating to radiation dose delivered to the heart and there is a need to develop dose-response relationships for radiation-related heart disease in this population.
The RTOG 0617 randomised trial reported an association between heart dose parameters (such as heart V5) and poorer overall survival [22,23]. Two lung cancer studies reported a relationship between MHD and coronary events. In one study of patients with no pre-existing cardiac disease a MHD > 10 Gy predicted major coronary events [24]. In another study of patients with pre-existing cardiac disease MHDs of 5 and 12 Gy predicted grade 3 cardiac event rates of 10% and 15% respectively [21].
The aim of this study is to present a systematic review of heart doses reported in in the modern era of lung cancer RT, including heart dose variation according to region irradiated, laterality, treatment modality and planning technique. It aims to summarise treatment strategies, such as motion management or use of particle therapies, which may lead to a reduction in heart dose. This work may guide the design of epidemiological and clinical studies investigating radiation-related heart toxicity in patients with lung cancer.

Study identification
EMBASE and SCOPUS electronic databases were searched for publications using the following terms in the title or abstract: 'dos* AND lung* AND cancer/carcinoma/tumor/tumour/ AND radi-ation/radiotherap' to retrieve all lung cancer studies reporting radiation dosimetry information published between January 1st 2013 and October 10th 2020.

Study eligibility criteria
Eligible studies were identified using the Preferred Reporting Items for Systematic Reviews guidelines ( Fig. 1) [25]. Studies reporting heart exposure from diagnostic imaging or from occupational/environmental radiation exposure and studies of lung cancer metastases located outside the thorax or of thoracic tumours that were not lung cancer e.g., thymoma and mesothelioma were excluded. Reviews, case studies and surveys were also excluded. All studies reporting any measure of whole heart or doses to cardiac substructures, the pericardium or the great vessels were eligible for inclusion in the study, and the treatment parameters associated with each heart dose reported within a study was described as a ''regimen". In total 1031 RT regimens in 281eligible studies were identified (Fig. 1, Table 1, Table S6).

Data extraction
For each eligible study the following variables were extracted for each regimen reported: author, year, country of first author, RT dose fractionation schedule, treatment modality, radiation modality, histology, location of primary tumour (laterality and lobar location), treatment planning technique, type of respiratory motion management used, cardiac delineation details, treatment planning algorithm used, number of CT plans per regimen included in the study and whether the treatment plan was delivered (Table S1, S6, Text S1). The tumour locations recorded included the lobar location of the lesion and whether the lesion was central, ultracentral or peripheral (Table S1). Regimens were categorised according to (1) treatment modality: stereotactic ablative body radiotherapy (SABR) or non-SABR RT (conventional or hypofractionated dose schedules, 1.5-3 Gy), (2) radiation modality: photon beam therapy or particle beam therapy (3) treatment planning technique: 3D-conformal RT; intensity modulated RT (static gantry IMRT, volumetric modulated arc therapy, helical tomotherapy, dynamic conformal arc therapy, MR-Linac); robotic driven delivery systems (cyberknife or X-knife); or particle beam therapy (protons or carbon ions) and (4) respiratory motion management: no respiratory motion management; non-active (Internal Target Volume approach, MidVentilation, MidPosition); or active (inspiration breath hold, expiration gating, inspiration gating, abdominal compression, respiratory tracking) (Table S1, Fig. 2). Dose optimiser objectives (the dose goal for the various structures and the priority of meeting each goal during the plan optimisation process) and dose volume constraints (DVCs) for the heart were extracted for IMRT and particle beam therapy regimens only. This was to determine the priority of the heart in inverse planning optimisation. For a thorough description of dose calculation algorithms reported in  1 Some studies contributed doses for several regimens and so contribute more than once. 2 Some regimens reported several dose measures e.g MHD and max heart dose. 3 Includes one regimen that was categorised as ultracentral, Four studies did not specify how central tumours were defined, two studies referred to RTOG 0236 guidelines i.e. within 2 cm of the proximal bronchial tree and one study referred to RTOG 0813 guidelines i.e. tumors within or touching the zone of the PBT, defined as a volume of 2 cm in all directions around the PBT, a tumor within 2 cm in all directions around the PBT and immediately adjacent to mediastinal or pericardial pleura (PTV touching the mediastinal pleura). 4 Includes all non-SABR regimens reporting a conventional or hypofractionated dose schedule. Dose per fraction of 1.5-3.0 Gy per fraction was reported in non-SABR regimens. 5 Non-SABR regimens were reported by tumour stage rather than location as mean heart dose and maximum heart dose was reported by lobar location in just 16%(64/392) regimens in (19/105) studies while information on tumour stage was available for 27%(107/392) regimens in34/105 studies. 6 This total represents 32 studies and 105 regimens also reporting whole heart doses and 6 studies and 25 regimens reporting substructure dose only. 7 This total represents the number of studies reporting some measure of whole heart and substructure dose.
studies please refer to Martino et al. [26]. Physical doses for the whole heart and doses to cardiac substructures, the pericardium and the great vessels were extracted from all regimens including volumetric and point doses. Doses reported relate to physical dose rather than the biologically effective dose to the heart as physical doses were reported and dose volume histograms were not available to allow conversion to EQD2.

Data analyses
Meta-analyses considered the average whole heart dose measures from the CT plans included for each regimen described in each study. Whole heart doses (average mean heart dose and average maximum heart dose) were compared according to laterality, lobar regions irradiated, treatment modality, radiation modality, treatment planning technique, and use of respiratory motion management techniques. Findings were reported separately for SABR and non-SABR regimens as the rationale for both techniques vary. Average mean heart dose and its 95% confidence interval was plotted along reported ranges. Under the fixed-effect model, the heterogeneities between and within SABR and non SABR regimens were tested by the method of variance-weighted least squares (VWLS). VWLS treats the estimated variance as if it were the true variance therefore it takes account different study sizes into heterogeneity test.

Results
In total, 978 regimens in 275 eligible studies were identified (Table 1, Table S6, Text S1-S2). 391 regimens in 98 studies were SABR studies and 587 regimens in 177 studies were non-SABR studies. Mean heart dose was the dose-parameter most reported  Heart doses varied according to location and disease stage ( Table 1). For SABR, MHD was higher in right-sided regimens (4.6 vs. 2.9 Gy) (Table S2). Information on lobar location of the primary lung lesion was available for 87/168 (54%) SABR regimens reporting mean heart dose, which was highest for central and left lower lobe lesions (6.3 Gy and 5.8 Gy respectively), and lowest for right and left upper lobe lesions (3.0 Gy and 1.8 Gy respectively). The highest maximum heart doses were for right middle lobe   (Table 1).
Heart doses varied according to radiation modality and treatment planning technique used (Fig. 3). For SABR regimens, 162/168 (96%) were delivered using photon beam therapy and 6/168 (4%) using particle beam therapy. MHDs were lower for particle beam therapy regimens (2.0 Gy vs. 4.1 Gy). There was no statistical difference between MHDs from various photon-planning techniques. Average mean heart doses were similar for 3DCRT (3.3 Gy (0.3-10.3)) and IMRT (4.6 Gy (0-32.4)), p = 1.0. For SABR studies, dose optimiser objectives were described for one study [27] while heart DVCs were described in only 25% (11/45) of IMRT studies and 38% (3/8) of particle beam therapy studies reporting mean heart dose and/or maximum heart dose (Table S3). Five studies referred to RT guidelines or protocols and nine studies described specific DVCs. The most common DVCs reported were maximum heart dose < 30 Gy and the volume of the heart receiving 46 Gy < 0.03 cc, each reported in two studies (Table S4). For non-SABR regimens 343/392 (87%) were delivered using photon beam therapy and 49/392 (13%) using particle beam therapy (Fig. 3). For photons average MHDs were similar between 3DCRT and IMRT 10.6 Gy (0-24.5) and 10.9 Gy (0-48.4) respectively. MHDs were lower for particle beam therapy (6.9 Gy). Optimiser dose objectives were described in 2 studies [28,29] while heart DVCs were described in only 37% (23/62) of IMRT studies and 21% (3/14) of particle therapy studies reporting mean and/or maximum heart dose ( Table 2). Three studies referred to guidelines or protocols and the remaining 23 studies described specific DVCs. Mean heart dose < 26 Gy was the common DVC reported in 4 studies.
Heart doses varied according to respiratory motion management (Fig. 4). For SABR, 136/168 regimens in 18/35 studies reported mean heart doses where respiratory motion management techniques were used. Active motion management techniques were reported in 83/136 (61%), non-active motion management techniques in 48/136 (35%) and no motion management was used in 5/136 (4%) of regimens. Mean heart doses reported using active motion management techniques were half those reported using non-active motion management strategies (2.4 vs. 5 Gy). Lowest mean heart doses were reported when inspiration breath hold was used (2 Gy (0.1-5.1)). For non-SABR 208/392 (54%) regimens in 63/105 (60%) studies reported mean heart doses where respiratory motion management techniques were used. Non-active motion management techniques were reported in 168/208 (81%), active motion management techniques in 19/208 (9%) and no motion management was used in 21/208(10%) regimens. Respiratory motion management use reduced exposure with MHD of 11.4 Gy, 9.3 Gy and 7.4 Gy reported for no motion management, non-active motion management and active motion management respectively.
Various contouring guidelines for the heart, cardiac substructures, pericardium, and great vessels were specified in 39/285 (13%) studies (Table S4). 19 studies referred to international protocols, 10 referred to published atlases while 10 gave anatomical landmarks for delineation (S4). 38 studies and 130 regimens reporting doses to cardiac substructures, the pericardium or the great vessels were identified (Table S5, Text S2) with heterogeneity in the range of structures and corresponding dose metrics reported. For SABR studies, D0.1 cc Aorta (the minimum dose received by 0.1 cc of the aorta) was most commonly reported in 25/90 (27.8%) regimens in 1/28(3.6%) studies. For non-SABR studies, the doses received by 67% and 33% of the pericardium were most commonly reported in 18/49 (36.7%) regimens in 2/18 (11.1%) studies (Table S5).

Discussion
This systematic review of heart doses in lung cancer RT regimens published between 2013 and 2020 demonstrates the heterogeneity of heart doses in clinical practice and reports a number of strategies that influence cardiac exposure systematically. MHD was reported in over half the studies identified, suggesting that our results truly reflect average cardiac exposure during 2013-2020. It was not possible to extend our analyses beyond MHD as other parameters were reported less frequently. As yet there are no international guidelines on dose reporting for the heart or substructures of the heart specific to lung cancer RT. Quantec specified mean heart dose (MHD) < 26 Gy as a dose volume constraint (DVC) for thoracic RT but this constraint was not confirmed from RT studies in patients with lung cancer [30]. Dose distribution is not homogenous within the complex structure of the heart [31] and a number of recent studies have suggested specific dose parameters beyond MHD associated with cardiac events that may be considered in dose reporting. These include V30 < 50% (volume of the heart receiving 30 Gy should be < 50%), V45 < 35% [32], V5 should be as low as possible [23,33] left anterior descending coronary artery (LADCA) V15 > 10% and left ventricle V15 < 1% [24]. The percentage of studies reporting, whole heart V30 and V45 was 20% and 5% respectively. LADCA V15 and left ventricle V15 were not reported in any of the studies identified. This may be because these parameters were only specified in studies published as recently as 2018 and 2020. Our study also demonstrates that cardiac contouring guidelines and atlases used vary between studies which contributes to dose reporting uncertainties. An international consensus on the use of cardiac contouring atlases and dose reporting may be warranted to provide recommendations for the heart specific to lung cancer RT in clinical practice.
As expected, heart radiation doses (physical doses) were much less for SABR than non-SABR regimens (MHD 4.0 (range < 0.1-32.4) Fig. 3. Average whole mean heart doses reported in lung cancer radiotherapy published 2013-2020 according to radiotherapy treatment modality, radiation modality and treatment planning technique used. Radiotherapy planning technique definitions are provided in Table S1. Abbreviations: Gy: gray; SE: standard error; CI: confidence interval; SABR: stereotactic ablative body radiotherapy; IMRT: intensity modulated radiotherapy; DCAT: dynamic conformal arc therapy. *Average of mean heart doses for reported regimens. yRange of mean heart doses for reported regimens. v 2 1 for linear trend test, v 2 n for heterogeneity test. All doses reported refer to physical doses.
Exposure of the Heart in Lung Cancer Radiation Therapy vs 10.3 Gy (range < 0.1-48.4)). For SABR, exposure was higher in central and lower lobe lesions (6.3 and 5.8 Gy respectively) compared to other locations. MHD was lowest for carbon ions (0.5 Gy) compared to other techniques. Active respiratory motion management reduced exposure (2.4 Gy versus 5.0 Gy). For non-SABR, MHDs were not significantly different between left and right-sided tumours. For non-SABR photon-based therapy MHDs were similar between IMRT and 3DCRT (10.9 Gy versus 10.6 Gy). The most common heart DVC used for inverse planned IMRT was MHD < 26 Gy. MHDs were lower with particle therapy (proton 7.0 Gy; carbon-ion 1.9 Gy) and active respiratory motion management significantly reduced exposure (7.4 Gy versus 9.3 Gy). Despite clear evidence of excess cardiac exposure in left-sided breast cancer RT [34] this was not observed in left-sided lung cancer RT. This suggests that a reliable indication of the effect of RT on the heart may not be obtained by comparing outcomes for patients with left-sided lung tumours to those with right-sided tumours. Lobar location may be a more significant indicator of prognosis with lower lobe tumours associated with poor prognosis in NSCLC [35,36]. In lung cancer, there has been a shift from 3DCRT to IMRT planning techniques and the use of particle beam therapy. IMRT was most frequently reported planning technique in lung cancer RT in this study except in complex cases where IMRT in the presence of multiple or larger lesions failed to meet dose volume constraints [32,37,38]and particle therapy was used. Despite wide use of inverse planning our results show that heart doses are similar from 3DCRT and IMRT regimens. This result is in contradiction with that of RTOG0617 showing that IMRT reduces heart doses [39]. The most common heart DVC reported was MHD < 26 Gy suggesting there may be scope to tighten heart DVCs. It may be possible to further prioritise the heart in plan optimisation. Only two studies emphasise heart dose as a clinical priority and non-co planar arcs were used for avoidance [40,41]. Studies emphasised using partial arcs to reduce contralateral lung dose [42][43][44] and planning optimisation to specifically reduce pneumonitis [45,46] and secondary breast cancer risk [46]. Particle therapy, specifically C-Ion, reduced mean and maximum heart doses in early stage SABR patients [37]. Active respiratory motion management, most commonly deep inspiration breath hold, resulted in lower heart doses and was commonly reported in SABR regimens in our study. This may be due to the increased tumour motion observed in earlystage NSCLC compared with locally advanced disease where a non-active approach could lead to the irradiation of excess healthy tissue. Non-active respiratory motion management was commonly used in non-SABR regimens and resulted in cardiac sparing compared to studies where no motion management was used. Overall, for non-SABR, an active approach resulted in lower cardiac exposures with average mean heart doses using expiration gating half those reported using the internal target volume non-active approach. Average mean heart doses were not reduced with breath-hold techniques specifically but these regimens involved patients where elective nodal irradiation was delivered and avoidance of the oesophagus was prioritised during optimisation [47].
Our study has several strengths. This is the first systematic review of heart doses in lung cancer RT. Data was extracted by an oncologist and radiation therapist and checked by a physicist. Table 2 Dose Volume Constraints for the heart described in IMRT and particle beam therapy non-SABR regimens reporting mean and maximum heart dose in lung cancer radiotherapy 2013-2020. There was a wide search strategy capturing relevant studies. Our study has a number of limitations. Doses reported relate to physical dose only. This is due to the nature of the linear quadratic model where calculating the EQD2 of the mean dose of a whole organ would lead to an incorrect estimate of the effective dose, especially in the presence of sharp dose gradients present in SABR or proton plans. Information on PTV volumes sizes was not always reported and could not be systematically extracted so it was not possible to identify the impact tumour size has on treatment planning technique chosen and associated cardiac exposure. In addition, information on clinical outcomes was limited in the studies identified and it was not possible to summarise cardiac exposures and clinical cardiac events. It is as yet unclear which regions of the heart are most sensitive to radiation induced injury and the dose thresholds to apply in the routine setting [48]. Pre-existing cardiac conditions may merit more stringent dose volume constraints [49] but inconsistency in reporting of cardiac endpoints and baseline cardiac risk mean that further validation is required [24]. The need to identify specific cardiac substructures and dose volume relationships to improve cardiac risk estimation is consistently cited [24]. More widespread reporting of cardiac dosimetry is essential, and this may be facilitated by developing international consensus guidelines around the topic to create awareness of what is needed to facilitate clinical research.
In conclusion, there is scope to reduce heart radiation doses in lung cancer RT. For photon based IMRT, the most common technique used in the clinic, more stringent planning optimisation objectives may reduce heart dose. Active respiratory motion management or particle therapy may also be considered where cardiac dose is high. There is an unmet need to understand the underlying mechanisms leading to RT-related cardiac toxicity and the impact on the sub-structures of the heart. Such information is vital in order to develop studies leading to the definition of more precise dose limits for the heart and substructures of the heart. In addition, consensus on planning objectives, contouring and DVCs for the heart are important objectives in order to validate more accurate dose volume relationships resulting in improved outcomes in patients with lung cancer.  Fig. 2 and Table S1. Abbreviations: Gy: gray; SE: standard error; CI: confidence interval; SABR: stereotactic ablative body radiotherapy. *Average of mean heart doses for reported regimens. yRange of mean heart doses for reported regimens. v 2 1 for linear trend test, v 2 n for heterogeneity test. All doses reported refer to physical doses.
Exposure of the Heart in Lung Cancer Radiation Therapy