Conformal Avoidance of Normal Organs at Risk by Perfusion-Modulated Dose Sculpting in Tumor Single-Dose Radiation Therapy

Abstract

Purpose:

Although 24 Gy single-dose radiation therapy (SDRT) renders >90% 5-year local relapse-free survival in human solid tumor lesions, SDRT delivery is not feasible in ~50% of oligometastatic lesions owing to interference by dose/volume constraints of a serial organ at risk (OAR). Conformal OAR avoidance is based on a hypothetical model positing that the recently described SDRT biology specifically permits volumetric subdivision of the SDRT dose, such that high-intensity vascular drivers of SDRT lethality, generated within a major tumor subvolume exposed to a high 24 Gy dose (high-dose planning target volume [PTV_HD]), would equilibrate SDRT signaling intensity throughout the tumor interstitial space, rendering bystander radiosensitization of a minor subvolume (perfusion-modulated dose sculpting PTV [PTV_PMDS]), dose-sculpted to meet a serial OAR dose/volume constraint. An engineered PTV_PMDS may thus yield tumor ablation despite PMDS dose reduction and conformally avoiding OAR exposure to a toxic dose.

Methods and Materials:

Dose fall-off within the PTV_PMDS penumbra of oligometastatic lesions was planned and delivered by intensity modulated inverse dose painting. SDRT- and SDRT-PMDS–treated lesions were followed with periodic positron emission tomography/computed tomography imaging to assess local tumor control.

Results:

Cumulative baseline 5-year local relapse rates of oligometastases treated with 24 Gy SDRT alone (8% relapses, n = 292) were similar in moderate PTV_PMDS dose-sculpted (23-18 Gy, n = 76, 11% relapses, P = .36) and extreme dose-sculpted (<18 Gy, n = 61, 14% relapses, P = .29) lesions, provided the major 24 Gy PTV_HD constituted ≥60% of the total PTV. In contrast, 28% of local relapses occurred in 26 extreme dose-sculpted PTV_PMDS lesions when PTV_HD constituted <60% of the total PTV (P = .004), suggesting a threshold for the PTV_PMDS bystander effect.

Conclusion:

The study provides compelling clinical support for the bystander radiosensitization hypothesis, rendering local cure of tumor lesions despite a ≥25% PTV_PMDS dose reduction of the 24 Gy PTV_HD dose, adapted to conformally meet OAR dose/volume constraints. The SDRT-PMDS approach thus provides a therapeutic resolution to otherwise radioablation-intractable oligometastatic disease.

Introduction

Recent phase 2 clinical trials reported that 24 Gy single-dose radiation therapy (SDRT) renders 92% actuarial 5-year local relapse-free survival (LRFS) in a wide range of human oligometastatic tumors, regardless of tumor type, size, or oligometastatic target organ.^1,2 However, SDRT was not feasible in multiple clinical settings because of the interference of dose/volume constraints of adjacent serial normal organs at risk (OARs).¹ Serial organs consist of discrete functional stem cell niches or other tissue-critical subunits that exhibit sensitivity to radiation in a dose-quantum binary mode.³ The cohort of serial organs commonly involved in oligometastatic settings (eg, central nervous system and major nerve trunks, gastrointestinal organs, the pulmonary bronchial tree, bladder wall) exhibit a threshold intolerance of point volume (≤0.035 cm³) to single exposure within the range of 14 to 18.5 Gy.⁴ Whereas baseline radiotherapeutic policy prioritizes prevention of critical OAR toxicity,^5,6 24 Gy–intractable oligometastatic lesions were treated in the phase 2 SDRT trial¹ with an alternative standard of care, nontoxic 3 × 9 Gy stereotactic body radiation therapy (SBRT). This treatment schedule was selected because of its acceptably low grade ≤2 neurologic toxicities when used in treatment of metastatic epidural tumors.⁷ Although this regimen was well tolerated, the 5-year actuarial LRFS was reduced to 38%.¹

The issue of restrictive proximity of OARs is a well-established limitation to tumor eradication by hypofractionated SABR, the current predominant radioablative technique in oligometastatic radiation therapy.^8,9 Ultrahigh-dose SABR schedules (3 × 18-20 Gy) render >90% actuarial 2- to 5-year LRFS of oligometastatic lesions,^10,11 albeit subject to the same OAR dose/volume restrictions as SDRT. Although reduced dose-intensity SABR results in suboptimal long-term LRFS,¹² recent literature has emphasized the need for comprehensive ablation of all clinically detectable lesions in early metastatic cancer to effect cancer cure.^9,13 Reduction of the SABR biological equivalent dose to levels compatible with tumor/OAR constraints is often substantial,^4,8,14-16 necessitating acceptance of increased risk of in-field local relapse.⁶ Alternatively, dose reduction can be used selectively in a minor tumor subvolume intersecting the planning risk volume (PRV) of the OAR, a technique termed simultaneous integrated protection.¹⁷ The International Commission on Radiation Units and Measurements (ICRU) Report 83 raised concerns regarding such approaches,⁶ which defy a baseline tenet of tumor cure by dose-fractionated radiation therapy.

Conceptually, dose fractionation is based on the notion that radiation tumor response is tumor cell autonomous, critically dependent on inherent tumor type–specific variations in repair fidelity of DNA double-strand breaks (DSBs),^18-20 with error-prone DSB repair rendering tumors radiosensitive whereas faithful repair yields radioresistance.^18-20 Given that slow proliferating normal tissues are systematically more effective in conducting faithful DSB repair than tumor tissues,²¹ a therapeutic ratio is reached. A corollary of this model is the widely accepted notion that every tumor cell within a tumor mass represents an independent target for radiation therapy, mandating delivery of a tumor type–specific tumoricidal dose to each cell to maximize tumor control probability.^5,6,22,23 Consistent with this notion, 2 recent studies using simultaneous integrated protection in oligometastatic tumors reported 12-month LRFS of 77% and 67%, respectively.^24,25 Hence, at present, the problem of SDRT and SABR implementation in the clinical setting of tumor/OAR dose conflicts remains unresolved.

The present study explores the hypothesis that in contrast to fractionated radiation therapy, newly discovered drivers of SDRT biology may specifically permit preplanned reduction of dose within a limited tumor subvolume to provide tumor cure with OAR conformal avoidance. The biologic mechanism of SDRT differs fundamentally from the mechanism driving fractionated radiation therapy²; it engages a synthetic dual-target mechanism that involves rapid endothelial acid sphingomyelinase (ASMase)-mediated vasoconstriction and global tumor hypoxia, disabling tumor cell homology-directed repair of SDRT-induced DSBs.² Ischemic hypoxia, generated within minutes, prevails throughout the interstitial tumor space and is sensed by parenchymal tumor cell mitochondrial complex III,²⁶ which, in response, releases reactive oxygen species,^2,26 triggering an adaptive oxidant SUMO stress response within tumor cells. One outcome of the SUMO stress response is depletion of nuclear chromatin-bound free SUMO3.² Whereas SUMO3 is required for enzymatic activation of multiple homology-directed DSB repair factors (ie, RAP80, BRCA1, RAD51) within repair foci,² loss-of-function homology-directed repair renders catastrophic DSB unrepair, chromosomal/micronuclei disarray, and massive tumor clonogen demise.² ASMase-mediated ischemic hypoxic injury appears rate-limiting for SDRT tumor cure^2,27; preventing SDRT-induced vasoconstriction with the endothelin-1 inhibitor BQ-123, or postirradiation scavenging of tumor cell reactive oxygen species, averts chromatin-bound SUMO3 depletion, homology-directed repair dysfunction, and SDRT tumor cure. Clinical studies using diffusion-weighted magnetic resonance imaging confirm that human oligometastatic lesions treated with 24 Gy SDRT, but not with hypofractionated 3 × 9 Gy SBRT, engage this vascular ischemic response in tumor ablation.²

Given these basic attributes of the SDRT model, we posited that the specific mechanism of high-dose SDRT enables dose sculpting of a sub-planning target volume (PTV) region of the tumor, engineered by volumetric intensity modulated dose painting^28,29 to render a localized PTV dose fall-off from 24 Gy to a dose complying with an OAR toxicity threshold. Dose reduction would locally generate only low-intensity ASMase-mediated hypoxia, insufficient to yield lethal DSB unrepair in tumor cells, thereby predisposing patients to in-field relapse. This circumstance might be mitigated by high-intensity ASMase signaling of SDRT acute hypoxia within the dominant body of 24 Gy–exposed PTV (termed here high-dose PTV [PTV_HD]). We posit that the 24 Gy–induced intensity of hypoxia might disperse across the entire tumor interstitial space due to equilibration of O₂ reduction, conferring sufficient DNA repair inactivation within the dose-sculpted tumor subvolume, thus providing a bystander effect to maintain the high tumor cure rate observed with full-tumor SDRT. Because the OAR contains an anatomically independent vascular/interstitial space, it would not be subject to this bystander effect. We have named this biology-driven, SDRT-specific model perfusion-modulated dose sculpting (PMDS), and the minor dose-sculpted subvolume designed to render OAR conformal avoidance is called the PTV_PMDS. Our preliminary clinical data, reported here, support the SDRT-PMDS hypothesis, confirming this technique is feasible and robust in effecting ablation of human oligometastatic disease within a wide range of dose sculpting, conformally avoiding toxic dose exposure to adjacent OARs.

Methods and Materials

Patients

The present study is an extension of a recently published institutional review board–approved clinical phase 2 study (clinicaltrials.gov NTC03543696) of 175 consecutive patients with oligometastatic disease undergoing radioablation.¹ The study permitted SDRT-PMDS treatment planning, at the discretion of the treating physician, when lesions could not be treated with 24 Gy SDRT alone because of dose/volume constraints of an adjacent OAR. The trial has closed to new patient accrual, and follow-up has continued to date, as per protocol rules. The latest update consists of a total of 598 lesions, treated with ablative intent as long as the patient maintained an oligometastatic status as previously defined¹; this represents an increase of 32 lesions¹ resulting from the interim sequential appearance of additional oligometastatic lesions in previously treated patients. Although this trial was a single-institution study involving a relatively small group of treating physicians, potential selection bias in exploring SDRT-PMDS versus referring OAR-conflicting lesions for treatment with 3 × 9 Gy SBRT cannot be excluded. Hence, this study was restricted to providing proof of principle that SDRT-PMDS enables OAR conformal avoidance without increased incidence in local relapse in the dose-sculpted PTV_PMDS. The current analysis is restricted to observations on clinical and therapeutic variables affecting local tumor control.

Treatment planning and delivery

As previously published,¹ all lesions were planned for radioablation with [¹⁸F]-fluorodexoyglucose positron emission tomography (PET)/CT and/or [⁶⁸Ga]-PSMA-PET/CT used to outline the gross tumor volume (GTV). Four-dimensional CT scanning was used in pulmonary mobile lesions to design an internal target volume according to the Radiation Therapy Oncology Group study 0236 guidelines. The PTV was obtained with a 3-mm isotropic margin added to the GTV or internal target volume. OARs were expanded with isotropic margins to produce PRVs as indicated by ICRU Report 83.⁶ Treatment planning was carried out with the Eclipse software (Varian Medical Systems, Palo Alto, CA) using 2 to 4 coplanar volumetric modulated arc therapy (VMAT) 6MV or 10MV flattening filter free beams and IMRT inverse dose painting.¹ Plans were calculated using heterogeneity corrections to ensure appropriate dose calculation. The D99 dose-volume metric was used to indicate the minimum absorbed dose by the PTV, purging the 1% outlier PTV voxels with the lowest absorbed dose, often located in a high-dose fall-off region at the periphery of the PTV, to prevent bias in the calculations of dose/volume effects on the risk of local relapse. Thus, the SDRT alone planning objectives were to achieve a PTV D99 ≥24 Gy plan normalization with an isotropic dose gradient of the PTV penumbra, complying to accepted OAR dose/volume constraints⁴ if present.

When the PTV penumbra did not comply with OAR dose/volume constraints, the SDRT-PMDS technique was used as a protocol option, reducing the PTV_PMDS as required to comply with the OAR constraints.⁴ Dose sculpting was achieved using dose painting by VMAT.²⁹ A dose fall-off was engineered regardless of beam energy at rate of up to 10% to 12%/mm by deploying ≥3 VMAT arcs,²⁹ assuring that the PTV_PMDS outer penumbra surface isodose interacting directly with the OAR did not exceed the threshold tolerance of the OAR. The major PTV subvolume that did not undergo dose sculpting (PTV_HD) was covered by a D99 ≥24 Gy. When neither baseline SDRT treatment plan nor the SDRT-PMDS technique was deemed to comply with the serial organ dose/volume constraints, treatment planning was diverted to the nontoxic 3 × 9 Gy hypofractionated SBRT protocol, as previously described.¹

All plans underwent strict quality assurance testing with a pretreatment dry-run using ArcCHECK (Sun Nuclear Corp., Melbourne, FL). Gamma analysis values at the 3%/3 mm ≥90% were considered acceptable. The EDGE/TrueBeam STx platforms (Varian Medical Systems) equipped with an ExactCouch (6° of freedom) system were used for treatment delivery. Treatment was started after accurate cone beam CT matching and target realignment via the 6° of freedom couch.

Endpoints

Local tumor response was assessed according to the PERCIST (PET Response in Solid Tumors) guidelines,³⁰ using PET/CT scans, at 3 and 6 months posttreatment and at 6-month intervals thereafter. Actuarial LRFS and cumulative incidence of local recurrences of treated lesions were calculated based on events occurring for each lesion independently and calculated from the date of the radioablative exposure. Acute and late toxicities were scored based on the National Institutes of Health Common Terminology Criteria for Adverse Events Guidelines, version 4.0.

Statistical analysis

Time to local failure was calculated from the day of treatment. GTV, baseline SUV_max, oligometastatic lesion histology, target site, use of systemic therapy before treatment or postablation, the dose that covers 95% of the PTV, the near minimum dose to the entire PTV (D99), the mean dose to the PTV and its standard deviation, and the PTV covered by the prescription dose (V99) were all tested as variables affecting local control. Actuarial LRFS was calculated by the Kaplan-Meier algorithm, and the cumulative incidence of local failures was reported as 1 – Kaplan-Meier. To exclude the possibility that death without recurrence constituted a competing risk affecting the incidence of LR, the Fine and Gray model was used and showed absence of a competing risk.³¹ Univariate analysis was performed to compare the association of relevant variables using the Cox proportional hazards regression method. A complete-case stepwise multivariable analysis model was set to estimate covariates that had a statistically significant level at α = 0.10 in the univariate mode, using the Cox proportional hazard model. Hazard ratios and 95% confidence intervals were obtained, and the level of statistical significance was set at < .05. χ² tests and t tests were used, respectively, to compare the distributions of categorical and continuous variables in the treatment subgroups. Statistical calculations were performed using the R software version 3.4.4, or the GraphPad Prism 7.0 software (Prism Inc, Reston, VA).

Results

Patients and lesions

Table 1 summarizes the characteristics of the oligometastatic lesions included in the present analysis. Although the primary goal of the study was to achieve oligometastatic ablation with full-dose 24 Gy SDRT, deployment of this dose to the entire PTV was feasible in only 292 of 598 (49%) lesions, and 24-Gy SDRT exposure was incompatible with serial OAR dose/volume constraints in the remainder. In 162 (27%) lesions SDRT-PMDS was used, entailing delivery of a D99 of 24 Gy to the major PTV_HD subvolume and a sculpted OAR-compatible dose to the minor PTV_PMDS; 3 × 9 Gy SBRT was used in 144 (24%) lesions.

Table 1.

Characterization of lesion variables by designated treatment

Lesions	All n (%)	SDRT n (%)	PMDS n (%)	SDRT vs PMDSP value	SBRT n (%)	PMDS vs SBRT P value
	598 (100)	292 (49)	162 (27)		144 (24)
Histology				<.001		.029
Prostate	143 (24)	84 (29)	36 (22)		23 (16)
NSCLC	108 (18)	56 (19)	21 (13)		31 (22)
CRC	86 (14)	43 (15)	24 (15)		19 (13)
Breast	82 (14)	26 (9)	37 (23)		19 (13)
Other	179 (30)	83 (28)	44 (27)		52 (36)
Lesion site				.06		<.001
Bone	193 (32)	94 (32)	61 (38)		38 (27)
Lymph nodes	212 (35)	81 (28)	47 (29)		84 (58)
Liver	47 (8)	26 (9)	16 (10)		5 (3)
Lung	105 (17)	70 (24)	26 (16)		9 (6)
Soft tissues	47 (8)	27 (9)	12 (7)		8 (6)
Lesion volume				<.001		.21
Median	5.1	4.8	6.1		7.5
Mean ± 1 SD	15.6 ± 37.8	7.9 ± 6.9	20.0 ± 36.9		27.6 ± 63.8
SUVmax				.002		.81
Median	9.2	6.1	7.3		10.2
Mean ± 1 SD	6.8 ± 8.4	8.1 ± 12.9	10.4 ± 10.2		7.5 ± 8.7
Adjuvant treatment				.76		.25
Yes	335 (56)	158 (54)	99 (61)		78 (54)
No	263 (44)	134 (46)	63 (39)		66 (46)

Abbreviations: adjuvant Tx = adjuvant systemic therapy; CRC = colorectal cancer; NSCLC = non-small cell lung cancer; PMDS = dose-sculpted SDRT coverage planned to a PTV_PMDS subvolume to comply with radiation tolerance of an adjacent normal organ at risk (rest of the tumor treated with 24 Gy); SBRT = hypofractionated 3 × 9 Gy stereotactic body radiation therapy schedule; SD = standard deviation; SDRT = 24 Gy single-dose radiation therapy covering the whole tumor volume.

Figure 1 shows a typical SDRT-PMDS treatment plan for a colorectal oligometastatic lesion in a pelvic lymph node interacting with the sigmoid colon PRV. Although 24 Gy to the entire PTV was not feasible because the outer penumbra surface isodose interacting with the OAR PRV exceeded the colonic dose/volume exposure restriction to 14.5 Gy (Fig. 1A), an alternative SDRT-PMDS treatment plan (Fig. 1B) complied with OAR constraints. Despite significant compromise in the PTV mean dose of this tumor lesion (16.5 Gy), potentially associated with a substantial risk of local relapse,²² this lesion was eradicated and remained free of in-field local relapse at 3 years post–SDRT-PMDS (Fig. E1).

Fig. 1.

Color-wash dose distribution of treatment plans to treat an fluorodeoxyglucose-avid left pelvic lymph node metastasis from a primary colorectal cancer interacting with the sigmoid colon. The positron emission tomography/computed tomography–derived gross tumor volume was expanded into planned target volume (PTV) by an isotropic 3-mm margin. (A) A clinically infeasible plan with 95% of the PTV of 24.0 Gy (D50% 24.5 Gy; D2% 25.5 Gy; D98 24.2 Gy), as the outer penumbra dose interacts with the sigmoid colon at a near-maximum dose (D2%) of 19.2 Gy, exceeding the permissible 14.5 Gy dose/volume constraint. (B) An alternative single-dose radiation therapy-PMDS treatment plan created to comply with the organ at risk (OAR) constraint. The overlap between the PTV and the sigmoid colon was used to generate the PTV_PMDS subvolume dose sculpted to meet the 14.5 Gy colonic tolerance restriction. Dose fall-off across the PTV_PMDS was engineered by volumetric modulated arc therapy inverse dose-painting, rendering a dose gradient of 42% within the distal 4 mm of the PTV_PMDS penumbra interacting with the contour of the OAR. The high-dose PTV has a D95 of 24.3 Gy (D50% 24.9 Gy; D2 25.6 Gy; D98 24.2 Gy). The PTV_PMDS shows a pronounced inhomogeneity (D50% 19.8 Gy; D2% 24.5 Gy; D98% 14.5 Gy), with the outer isodose interacting with the OAR fulfilling the sigmoid colon D_max constraint of 14.5 Gy, thereby attenuating the radiation dose to the colonic wall to a nontoxic dose level.

Toxicity did not represent overall a significant concern in this series, irrespective of whether patients received full-dose 24 Gy SDRT, SDRT-PMDS, or 3 × 9 Gy SBRT. A total of 55 of 175 study patients (35%) experienced limited grade 1 acute toxicity not requiring symptomatic therapy, and no grade ≥2 acute or late toxicities were observed within a follow-up period of 3.0 to 78.9 (median 37.7) months across the entire patient population. Of note, patients with oligometastatic lesions at high risk of vertebral compression fractures after 24 Gy SDRT, such as baseline vertebral compression fractures, significant (>50%) lytic tumor burden per vertebra, and/or spinal instability neoplastic scores indicating an unstable spine, were not accrued to the present trial and were managed primarily with vertebroplasty.

SDRT-PMDS yields sustained local control of human oligometastatic lesions

Figure 2 shows the cumulative 5-year actuarial incidence of local relapses in lesions treated with 24 Gy SDRT to the whole tumor (no PMDS required) yielding an 8% local recurrence rate, likely associated with uncertainties in treatment planning or with mitigation of subdiaphragmatic mobile targets (ie, tumor lesions in the liver, adrenals).¹ When SDRT-PMDS was used, lesions exhibited a moderate increase in cumulative local relapse rate to 15% by 5 years, albeit not reaching statistical significance compared with the SDRT alone group (Fig. 2; P = .06). The 3 × 9 Gy SBRT cohort had a 46% rate of local recurrence (P < .001). The SDRT, SDRT-PMDS, and SBRT cohorts significantly differed by histology and oligometastatic target organ (Table 1). However, recently published data indicated that oligometastatic radioablation was independent of histologic tumor type or target organ,¹ indicating that the significance of histologic and target organ variations observed here represent outcomes of nonrandomized sampling.

Fig. 2.

Cumulative 5-year actuarial incidence of in-field local failure of oligometastatic lesions treated with ablative radiation therapy. Single-dose radiation therapy (SDRT) designates lesions with a prescribed single dose of 24 Gy covering the whole planned target volume (PTV) (PTV D99 ≥24 Gy). Perfusion-modulated dose sculpting (PMDS) indicates SDRT-PMDS treated lesions reducing the dose coverage in the PTV_PMDS to comply with the organ at risk constraints. When the PMDS plan failed to meet the organ-at-risk tolerance criteria, a hypofractionated 3 × 9 Gy stereotactic body radiation therapy scheme was used, designated here as single-dose radiation therapy. Actuarial analysis was used to calculate cumulative incidence of local failure, and statistical significance of treatment outcomes by treatment mode was calculated using the log-rank test.

Furthermore, recent studies indicated that tumor PET/CT-derived metrics, such as large tumor volumes and high metabolic SUV_max values, predicted poor prognostic phenotypes.¹ Table 1 shows that the SDRT-PMDS and SBRT cohorts did not differ statistically in mean GTV or in PET SUV_max values (P = .21 and P = .81, respectively), although both were statistically increased (P < .001 and P = .02, respectively) compared with baseline mean values of the 24 Gy SDRT cohort. Nonetheless, the local relapse rate in the 3 × 9 Gy SBRT cohort was 3-fold higher than in the SDRT-PMDS treated lesions, and the local relapse rates in the SDRT and SDRT-PMDS groups did not differ statistically (Fig. 2). This observation is consistent with the notion that the SDRT-based approaches and the 3 × 9 Gy hypofractionated SBRT operate distinct biologic mechanisms of action in tumor ablation.²

Further univariate analysis of the SDRT and the SDRT-PMDS cohorts using the Cox proportional hazards regression test (Table 2) revealed that colorectal histology and lung and soft tissue targets, as well as lesion volume, were significant variables affecting local recurrence probability. However, none of these effector variables were significant in the multivariate analysis (Table 2), although the PTV_PMDS dose/volume coverage parameters in a subgroup of extreme dose-sculpted SDRT-PMDS lesions indicated statistical significance in promoting in-field local recurrences.

Table 2.

Univariate and multivariate analysis of Cox proportional HR of relevant tumor and treatment variables with local recurrence after oligometastatic tumor radioablation

	Univariate				Multivariate
		95% CI				95% CI
	HR	Lower	Upper	P	HR	Lower	Upper	P
Histology
Other histology (ref.)
Breast	0.298	0.065	1.362	.118	4.4991	0.042	1.155	.074
Colo-rectal cancer	2.681	1.158	6.209	.021	0.4121	0.962	6.117	.060
NSCLC	1.716	0.713	4.128	.228	0.5005	0.753	5.297	.164
Prostate	0.285	0.078	1.035	.056	2.7172	0.091	1.485	.160
Lesion site
Bone (ref.)
Liver	2.446	0.632	9.463	.195	1.9025	0.096	2.853	.456
Lymph nodes	1.462	0.544	3.925	.451	0.8321	0.394	3.661	.746
Lung	2.825	1.095	7.290	.032	0.5656	0.593	5.262	.306
Soft tissues	4.644	1.625	13.269	.004	0.3506	0.902	9.018	.074
Volume
GTV (cm3) continuous	1.009	1.002	1.016	.015	0.9910	0.998	1.020	.091
Baseline SUVmax
SUVmax continuous	1.011	0.981	1.043	.475	0.9815	0.979	1.060	.352
Systemic therapy
Adjuvant systemic therapy	1.171	0.607	2.258	.637	0.1599	0.832	46.980	.075
PTV coverage (D99%)
D99% ≥23 Gy (ref.)
PMDS<23-18 Gy	1.865	0.979	3.555	.058	0.5813	0.795	3.719	.168
PMDS<18 Gy V99 ≥60%	1.008	0.241	4.216	.992	1.3043	0.144	4.062	.755
PMDS<18 Gy V99 <60%	3.361	1.304	8.664	.012	0.3137	1.072	9.470	.037

Abbreviations: CI = confidence interval; GTV = gross tumor volume; HR = hazard ratio; NSCLC = non-small cell lung cancer; PMDS = perfusion-modulated dose sculpting; PTV = planning target volume; SUV = standardized uptake volume.

Dose/volume effects on SDRT-PMDS local tumor control

To further explore PMDS dose/volume settings that might promote in-field local recurrence, we defined 2 PTV_PMDS subgroups: a moderate PMDS group, dose-sculpted to generate a dose fall-off to within the range of 23 to 18 Gy, and an extreme subgroup dose-sculpted to <18 Gy. Baseline mean PTV dose in the 292 lesions treated with SDRT alone was 24.2 ± 0.3 Gy, which, as previously discussed, was associated with a local recurrence rate of 8% (Fig. 2). In comparison, although the mean PTV dose in the 162 SDRT-PMDS lesions was reduced by about 30% (16.6 ± 4.6 Gy; P < .001), the dose-sculpted lesions exhibited nonetheless no significant difference in local recurrence rate (Fig. 2; P = .06), defying a basic tenet of clinical radiobiology that asserts reduction in mean tumor dose is invariably associated with reduced tumor control.^5,6,22 Further detailed analysis of this approach revealed that moderate SDRT-PMDS in 76 lesions treated to a mean D99 of 20.5 ± 1.2 Gy (P < .0001 vs SDRT alone; Fig. 3A) yielded an 11% 5-year cumulative local recurrence rate (Fig. 3B; P = .36 vs SDRT), and extreme SDRT-PMDS in 86 lesions treated to a mean D99 of 13.7 ± 3.8 Gy (P < .0001 vs the moderate PTV_PMDS group; Fig. 3A) yielded significantly increased 18% cumulative local recurrence (Fig. 3B; P = .02).

Fig. 3.

Dosimetric characterization by local control and local failure of oligometastatic lesions treated with single-dose radiation therapy (SDRT) alone or with SDRT–perfusion-modulated dose sculpting (PMDS). No PMDS designates treatment with 24 Gy SDRT (D99) to the whole planned target volume (PTV). Moderate PMDS designates lesions dose sculpted to generate a PTV_PMDS D99 fall-off to within the range of 23 to 18 Gy, and extreme PMSD designates PTV_PMDS dose sculpted to <18 Gy. (A) Scatter plots of local control and local relapse by treatment modality. Each dot represents a treated lesion. (B) Cumulative 5-year actuarial incidence of local relapse by treatment modality. The log rank analysis indicates moderate PMDS local failure does not differ significantly from no PMDS (P = .36), and for extreme PMDS the analysis indicates P = .02.

To explore how volumetric variations of PTV_HD, the posited generator of PMDS bystander effects, might affect local recurrence in the extreme dose-sculpted subgroup, we examined PTV_HD volume as a fraction of total tumor PTV (this percentage is termed V_HD). Mean V_HD in the moderate SDRT-PMDS group was 78.8% ± 26.0%, compared with 68.7% ± 30.1% for extreme SDRT-PMDS lesions (P <.001). Figure 4A shows that variations in V_HD correlated with posttreatment incidence of local recurrence. Analysis of the patterns of local recurrence in the SDRT-PMDS subgroups by V_HD indicated that in the extreme SDRT-PMDS category 5 of 11 (45%) relapses occurred in lesions with V_HD <60% (Fig. 4A). With use of a V_HD cutoff value at 60%, Figure 4B shows that extreme SDRT-PMDS lesions using V_HD ≥60% (61 lesions) had a 14% cumulative incidence of local recurrences, not statistically different from the 24 Gy SDRT control group (Fig. 4B, P = .29); 25 extreme SDRT-PMDS lesions using V_HD <60% had a significantly increased cumulative 28% rate of local recurrence (Fig. 4B, P =. 04). Furthermore, the 5 lesions that relapsed locally in the V_HD <60% category were phenotypically small in size, exhibiting a mean GTV of 5.1 ± 7.9 cm³ (median 1.6 cm³), significantly reduced compared with the 32.6 ± 8.7 cm³ (median 11.0 cm³, P = .01) of the 11 lesions exhibiting sustained local control.

Fig. 4.

Volumetric characterization by local control and local failure (LF) of the oligometastatic lesions treated with single-dose radiation therapy (SDRT) alone or with SDRT-PMDS. The volumetric unit V_HD designates the ratio of planned target volume (PTV) covering the major tumor subvolume with the 24 Gy dose (PTV_HD) as a percentage of total tumor PTV (PTV_HD + PTV_PMDS). (A) Scatter plots of local control and local relapse by V_HD. Each dot represents a treated lesion. (B) Cumulative 5-year actuarial incidence of local relapse by V_HD. Log rank analysis indicates that incidence of LF for extreme PMDS with V_HD ≥60% LF did not differ significantly from no PMDS (P = .29), and it was significant for extreme PMDS with V_HD <60% (P = .004).

In sum, the data indicate that the SDRT-PMDS approach is effective in rendering sustained LRFS, which is not statistically different from the ablative potential of full dose 24 Gy SDRT, except for in a minor subset of small tumor lesions (<2 cm³) using a V_HD <60% to provide extreme SDRT-PMDS dose sculpting to <18 Gy.

Discussion

The present studies define the dose/volume parameters of the SDRT-PMDS algorithm in the design of volumetric-specific PTV dose sculpting, providing proof of principle that the biology-driven SDRT-PMDS technique enables OAR conformal avoidance without increasing the risk of tumor local relapse despite the generic PTV_PMDS dose sculpting. Although this paradigm defies dogma of classical radiation therapy, the outcome data provide evidence that SDRT-PMDS is feasible in a wide range of tumor/OAR clinical settings, rendering robust therapeutic effectiveness in tumor cure despite PTV_PMDS dose sculpting to as low as a D99 of ≤14 Gy. Further differentiation of SDRT-PMDS subtleties would require a significantly higher enrollment in different settings to address clinically relevant issues, such as our finding that extreme SDRT-PMDS fails in small GTV lesions entailing a V_HD of <60%, whereas in larger GTV lesions extreme SDRT-PMDS renders a robust therapeutic outcome regardless of the percent V_HD. Nonetheless, our data support the notion that the SDRT-PMDS technique resolves a frequent limitation to SDRT application, converting intractable tumor/OAR clinical settings into SDRT tumor cure. Preservation of the whole-PTV 24 Gy LRFS level despite a 30% decrease in the tumor mean PTV dose in PMDS-treated lesions and the demonstration that extreme PTV_PMDS dose sculpting critically depends on the PTD_HD constituting ≥60% of the total tumor PTV provide compelling support for the bystander radiosensitization hypothesis. Nonetheless, proof of the biological mechanism mediating the high success of SDRT-PMDS still requires formal experimental investigation.

The low local recurrence rates observed under PTV_PMDS dose sculpting are remarkable in view of the high local recurrence observed when SDRT dose to the entire tumor volume is reduced. An early SDRT phase 1 dose escalation study reported a 25% rate of 3-year LRFS in oligometastatic lesions treated with full PTV coverage of 18 to 20 Gy, 69% with 21 to 22 Gy, and 82% after 23 to 24 Gy.³² Similarly, Yamada et al³³ reported SDRT outcome data in 811 oligometastatic lesions to the spine, demonstrating a cumulative 4-year local recurrence of 20% with low-dose SDRT (median 16.4 Gy; range, 14-17 Gy), compared with 2.1% with high-dose SDRT (median 22.4 Gy; range, 17-28 Gy).³³ Whether increased rescue of the PTV_PMDS can be achieved by treating the PTV_HD with doses exceeding 24 Gy remains unknown. In this context, clinical studies using 26 to 40 Gy SDRT alone^34-36 have not improved tumor local relapse-free survival during the 5-year actuarial rate of 92% reported with 24 Gy.¹

The SDRT-PMDS approach was developed empirically at the patient bedside, driven by emerging discoveries of the function of new drivers of tumor lethality in experimental tumor SDRT. Perhaps the first turning point was the discovery that dose-dependent intensity of SDRT-induced endothelial ASMase signaling, rather than sheer number of tumor cell DSBs, determines rate of SDRT tumor cure.²⁷ Although the maximal capacity of human endothelial ASMase in mediating SDRT-induced microvascular dysfunction is quantitatively unknown and may be saturated at about 24 Gy, ASMase overexpression via gene therapy selectively targeting tumor neoangiogenic endothelium in preclinical models provides a rationale for enhancing the single-dose therapeutic ratio.²⁷ The asmase gene therapy experiments used the adenoviral vector Ad5H2E-mVEGFR2-ASMase, which uses a pre-proendothelin promoter only active in dividing tumor neoangiogenic endothelium to overexpress ASMase.²⁷ This asmase gene transduction technique enabled 25% de-escalation of the SDRT ablative dose in MCA/129 fibrosarcomas, rendering in tumors exposed to 14 Gy an iso-cure response normally produced by 20 Gy SDRT.²⁷ Whereas ionizing radiation induces a constant number of DSBs/Gy across all mammalian cells,³⁷ these dose de-escalation data suggest that SDRT normally overproduces DSBs relative to the critical number required for tumor ablation and that dose-dependent intensity of ASMase-mediated microvascular vasoconstriction and resulting acute ischemic hypoxia are the rate-limiting variables in effecting tumor ablation.^2,27

The latter experiment was critical in formulating the PMDS working hypothesis, as genetic transduction of <25% of the neo-angiogenic vasculature was sufficient to yield tumor radiosensitization (Kolesnick, unpublished). This observation suggests that SDRT-activated ASMase itself and/or the consequent acute ischemic hypoxia equilibrate SDRT lethal signaling throughout the tumor interstitial space. Consistent with this notion, activated endothelial ASMase was reported to be secreted into tissue interstitial space^38,39 and into the serum of SDRT-treated, but not 3 × 9 Gy SBRT-exposed, mice and human patients (Higginson and Kolesnick, unpublished). Hence, it is reasonable to assume the SDRT-induced ischemic hypoxia, or other SDRT prolethal intermediary drivers, might access transinterstitially a dose-sculpted low ASMase-activated, local relapse–prone PTV_PMDS, conferring PTV_PMDS bystander radiosensitization and local tumor cure. Although clinical data are consistent with this hypothesis, there is as of yet no direct validation of the SDRT-PMDS mechanism in experimental animal models. Future studies will also need to address changes in the tumor immune microenvironment by SDRT, SDRT-PMDS, or SBRT in settings of systemic immunotherapy or potential abscopal immune responses. However, such studies are beyond the scope of the current trial.

Conclusions

We believe SDRT-PMDS represents a substantive advancement in the treatment of oligometastatic cancer. Recent prospective phase 2 randomized studies have reported that ablative consolidation of oligometastatic lesions with SABR improves progression-free survival compared with standard-of-care palliative therapy,^40-43 with the SABR-COMET trial reporting approximately doubling of both progression-free survival and overall survival.⁴² These observations were interpreted as hypothetically indicating that comprehensive oligometastatic lesion ablation may affect the natural history of the oligometastatic syndrome, attenuating metastatogenic conversion of microscopic tumor foci.⁴³ Hence, we contend that comprehensive synchronous and metachronous oligometastatic lesion ablation, disregarding the barrier of ≤5 synchronous lesions as an indication for tumor ablation, needs to be explored for its role in the cure of oligometastatic cancer. Consistent with this notion, the 24 Gy SDRT experience in ablation of oligometastatic lesions reported that 42% of the patients exhibited 1 to 6 sequential bouts of new oligometastatic (≤5) lesions, subject at each such event to lesion ablation by 24 Gy SDRT.¹ In some instances up to 20 cumulative lesions per patient were treated without evidence of grade > 1 toxicity.¹ The actuarial 5-year disease-free survival in patients receiving 2 to 6 cycles of ablative 24 Gy SDRT was 56%, compared with 20% in patients treated with 1 course of metastasis-directed 24 Gy SDRT (P < .0001).¹ It is within this context that we regard the SDRT-PMDS approach as progress in the SDRT state of the art, significantly expanding the catchment of many, if not the majority, of the ~50% oligometastatic lesions previously considered SDRT-intractable. Multiple clinically relevant issues, such as the lack of an indication to use extreme PMDS in low-GTV tumors involving a V_HD of <60%, remain to be resolved in defining SDRT-PMDS as a new standard of care. Nonetheless, we foresee the deployment of this approach as a step forward in the cure of early metastatic human cancer.