Esophageal Circumferential Dose-Length Histogram Parameters to Predict Dysphagia in Patients Receiving Thoracic Radiation Therapy: A Single-Institution Experience

Abstract

Purpose

Chemoradiation for locally advanced non-small cell lung cancer can cause severe esophagitis. Techniques to spare the contralateral esophagus may mitigate toxicity, but traditional dose-volume histograms (DVH) do not capture the degree of circumferential irradiation. We evaluated dose-length histogram (DLH) parameters as predictors of dysphagia compared with DVH metrics.

Methods

We retrospectively reviewed patients treated with definitive thoracic radiation therapy from 2019 to 2023. Descriptive statistics described the cohort. Clinician-reported (National Cancer Institute Common Terminology Criteria for Adverse Events [CTCAE] v.4) and patient-reported outcomes (PRO)-CTCAE v.1 dysphagia within 120 days of treatment start were collected prospectively. The McNemar test compared dichotomized scores. The length of esophageal full-circumferential (L_Fdose) and partial-circumferential irradiation (L_Pdose) was defined as the length of the esophagus with ≥90% and ≥50% circumference exposure to threshold radiation doses, respectively. Spearman correlation examined relationships between L_Fdose, L_Pdose, and volumetric (V_dose) parameters. Associations between L_Fdose, L_Pdose, and V_dose, and grade ≥2 dysphagia were evaluated using univariate logistic regression. Likelihood ratio tests assessed model fit.

Results

Of 107 patients, 86.9% (93) had non-small cell lung cancer, all received ≥60 Gy (median, 63 Gy; range, 60-70 Gy), and 94.4% (101) received concurrent chemotherapy. Patients and physicians reported rates of grade ≥2 dysphagia differently: 17 (15.9%) and 6 (5.6%), respectively (P = .0015). Each 0.5 cm increase in the length of partial-circumference esophagus receiving ≥55 Gy (L_P55) and 60 Gy (L_P60) resulted in increased odds of PRO-CTCAE dysphagia by 8% and 9%, respectively. Each 0.5 cm increase in full-circumference esophagus receiving ≥60 Gy (L_F60) resulted in 11% increased odds of PRO-CTCAE dysphagia. Esophageal DLH parameters, L_P55 and L_P60, correlated with the esophageal V₆₀ volumetric parameter strongly (ρ = 0.751 and 0.729, respectively). No DVH or DLH parameter predicted grade ≥2 CTCAE dysphagia.

Conclusion

Esophageal DLH metrics assessing partial- or full-circumferential esophageal irradiation, specifically L_P55, L_P60, and L_F60, are associated with patient-reported dysphagia and complement traditional DVH parameters.

Introduction

Chemoradiation (chemoRT) followed by immunotherapy is the definitive treatment for patients with inoperable locally advanced non-small cell lung cancer (NSCLC). However, treatment confers a risk of severe esophagitis. Esophagitis is the leading cause of radiation-associated hospitalization among patients with lung cancer, with higher-grade esophageal toxicity associated with a 54% increase in the risk of death as demonstrated in randomized radiation therapy (RT) dose-escalation trials.^1,2 Thus, strategies are needed to reduce the incidence and severity of esophagitis.

Traditionally, esophageal toxicity has been predicted using dose-volume histogram (DVH) metrics to quantify the mean RT dose (D_mean) to the entire esophagus and the percent volume of the esophagus exposed to a threshold dose (V_dose). In the acute setting, radiation-associated esophageal toxicity presents as dysphagia. A meta-analysis of patients receiving chemoRT from 1993 to 2011 showed that the esophageal V₆₀ is a predictor of grade ≥2 and ≥3 acute esophagitis, with a V₆₀ > 17% conferring a high risk of radiation esophagitis.³ This volume-based metric that informs treatment planning of large phase stage III lung cancer trials is based on data from patients treated with 3-dimensional conformal RT and intensity modulated RT (IMRT) techniques.

The contralateral esophagus sparing technique (CEST) limits high-dose RT to the full circumference of the esophagus by employing the contralateral esophagus wall contoured as an avoidance structure for inverse RT planning.⁴ The National Comprehensive Cancer Network guidelines currently promote the CEST; however, there may be significant interobserver variability in contouring the contralateral esophagus substructure, and there is no standardized metric that quantifies partial-circumferential esophagus irradiation.⁵

Furthermore, there is evidence that clinician-assessed dysphagia underestimates toxicities experienced by patients because patient-reported outcome measures are more likely to report moderate difficulty swallowing earlier and more often than clinician assessments.⁶ Consequently, RT planning techniques and dosimetric constraints that incorporate knowledge of avoiding full-circumferential esophageal irradiation or contralateral-esophageal sparing are essential if we are to reduce the risk of symptomatic dysphagia and improve treatment-related quality of life with modern-era RT.

We previously described a dose-length histogram (DLH), a metric that quantifies the extent of partial- and full-circumferential esophagus irradiation, where the length of the esophagus receiving the threshold radiation dose to at least 50% and 90% of its circumference can be quantified.⁷ In this single-institution retrospective review of patients with thoracic malignancies treated with definitive doses of RT, we will evaluate if DLH parameters, quantifying avoidance of circumferential esophagus irradiation, are associated with patient-reported and/or clinical-graded dysphagia. In addition, we describe DLH parameters in relation to those of traditional DVH.

Methods

Patient identification and toxicity assessments

We identified patients with primary or secondary intrathoracic malignancies treated with conventionally fractionated RT (1.8-2.0 Gy/day) at Johns Hopkins from 2019 to 2023. These patients were treated using intensity-modulated or volumetric-modulated arc therapy (IMRT) with daily cone-beam image verification. The esophagus was delineated from the postcricoid esophagus to the gastroesophageal junction. Patients were excluded if they underwent adaptive replanning or if esophageal contours were unavailable.

Demographic, clinical, and treatment data were obtained from medical records. Data included age, sex, Eastern Cooperative Oncology Group performance status, histology, stage, total dose, fractionation, and the receipt of concurrent chemotherapy. Dosimetric data retrieved from archived electronic RT plans included baseline esophageal length, volume, mean dose, and V₆₀ parameters.

As the standard of care during weekly on-treatment visits, the treating physician graded dysphagia using National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE, version 4.0). Patients prospectively reported the severity of dysphagia using the patient-reported outcomes CTCAE (PRO-CTCAE, version 1.0) dysphagia assessment. The severity of dysphagia was determined by the highest score recorded within 120 days of initiating RT.

DLH metrics and evaluation

The DLH metrics were calculated by multiplying the number of contoured esophagus subvolumes that received partial- or full-circumferential dose exposures by the image slice thickness (Fig. E1). Specifically, the length of esophageal full- (L_Fdose) and partial-circumferential (L_Pdose) irradiation to a threshold dose were defined as the length of the esophagus with ≥90% and ≥50% circumference exposure to threshold radiation doses, respectively. Esophageal V₆₀ is defined as the percent volume of the esophagus receiving at least 60 Gy. Total esophageal length was calculated by multiplying the number of axial computed tomography slices with an esophagus contour by the computed tomography slice thickness. DLH parameters were calculated as previously described.⁷

Statistical analysis

Descriptive statistics described the cohort. We applied the McNemar test to examine the relationship between dichotomized grade 0 to 1 versus ≥2 CTCAE and PRO-CTCAE dysphagia scores. We evaluated associations between L_Fdose, L_Pdose, and the development of grade ≥2 CTCAE and PRO-CTCAE dysphagia using univariate logistic regression and restricted cubic spline in logistic regression for various threshold doses of L_Fdose and L_Pdose in increments of 5 Gy. The odds ratio for the CTCAE or PRO-CTCAE dysphagia was generated in relation to 0.5 cm increments of exposure to the DLH parameter and in relation to 2% increments of exposure to the DVH parameter. We used the Spearman correlation to examine the relationship between esophageal DVH parameters (L_Pdose and L_Fdose) and established DVH parameters (V_dose). A likelihood ratio test compared various esophageal DLH and DVH parameters to the null model to assess the metric that would allow the best model fit for dysphagia, with higher values indicating stronger predictive power.

Results

Patient characteristics

We identified 141 patients who received RT without adaptive replanning between 2019 and 2023. CTCAE and PRO-CTCAE dysphagia scores were unavailable for 14 patients. DLH calculations were available for 107 patients who were included in the final analysis. The median age was 66 years (range, 49-85). The majority of the patients were male (56.1%), had Eastern Cooperative Oncology Group performance status of 0 to 1 (90.7%), had NSCLC histology (86.9%), and received concurrent chemotherapy (94.4%). Among the 93 patients with NSCLC, 12 (12.9%) were stage II, 56 (60.2%) stage III, 17 (18.3%) stage IV, and 8 (8.6%) had recurrent disease. The median total dose delivered was 63 Gy (range, 60-70 Gy). See Table 1.

Table 1.

Patient characteristics

Patient characteristics	All (n = 107)	Patients with grade ≥2 PRO-CTCAE dysphagia (n=17)	Patients with grade ≥ 2 CTCAE dysphagia (n =6)
Median age, y (range)	66 (49-85)	66 (55-74)	67 (55-72)
Sex
Male	60 (56.1%)	10 (58.8%)	3 (50.0%)
Female	47 (43.9%)	7 (41.2%)	3 (50.0%)
ECOG performance status
0	26 (24.3%)	1 (5.9%)	0 (0%)
1	71 (66.4%)	13 (76.5%)	6 (100%)
2	8 (7.5%)	2 (11.8%)	0 (0%)
3	2 (1.9%)	1 (5.9%)	0 (0%)
Histology
NSCLC	93 (86.9%)	16 (94.1%)	6 (100%)
SCLC	3 (2.8%)	0 (0%)	0 (0%)
Other lung histology	3 (2.8%)	1 (5.9%)	0 (0%)
Solitary fibrous tumor	2 (1.9%)	0 (0%)	0 (0%)
Pleural sarcoma	1 (0.9%)	0 (0%)	0 (0%)
Mediastinal metastasis from non-lung primary	5 (4.7%)	0 (0%)	0 (0%)
NSCLC
Stage II	12 (12.9%)	1 (6.3%)	1 (16.7%)
Stage III	56 (60.2%)	13 (81.2%)	4 (66.6%)
Stage IV	17 (18.3%)	2 (12.5%)	1 (16.7%)
Recurrent	8 (8.6%)	0 (0%)	0 (0%)
Median total dose (range)	63 Gy (60-70)	63 Gy (60-66)	63 Gy (60-63)
Dose per fraction
1.8 Gy	59 (55.1%)	12 (70.6%)	5 (83.3%)
2.0 Gy	48 (44.9%)	5 (29.4%)	1 (16.7%)
Concurrent chemotherapy
Yes	101 (94.4%)	16 (94.1%)	6 (100%)
No	6 (5.6%)	1 (5.9%)	0 (0%)
Weight loss
0%-5%	72 (67.3%)	8 (47.1%)	5 (83.3%)
5%-10%	27 (25.2%)	7 (41.2%)	1 (16.7%)
10%-20%	8 (7.5%)	2 (11.8%)	0 (0%)

Abbreviations: CTCAE = National Cancer Institute Common Terminology Criteria for Adverse Events; ECOG = Eastern Cooperative Oncology Group; NSCLC = non-small cell lung cancer; PRO-CTCAE = patient-reported outcomes National Cancer Institute Common Terminology Criteria for Adverse Events; SCLC = small cell lung cancer.

The median esophageal length contoured was 23.4 cm (range, 18.0-30.9 cm) with a median volume of 48.9 cm³ (range, 10.3-102 cm³). The median esophageal V₆₀ was 6.40% (range, 0-48.4%). The median L_P55, L_P60, and L_F60 were 1.50 cm (range, 0-15.0 cm), 0 cm (range, 0-14.4 cm), and 0 cm (range, 0-11.1 cm), respectively. See Table 2.

Table 2.

Esophageal dosimetric outcomes

Esophageal dosimetric parameters	All patients (n = 107)	Patients with grade ≥2 PRO-CTCAE dysphagia (n = 17)	Patients with grade ≥2 CTCAE dysphagia (n = 6)
Median length (range)	23.4 cm (18.0-30.9)	24.0 cm (18.0-30.9)	24.3 cm (19.2-28.8)
Median volume (range)	48.9 cm3 (10.3-102)	51.9 cm3 (30.2-72.5)	61.7 cm3 (35.4-70.7)
Median V60 (range)	6.4% (0-48.4)	14.6% (0-48.4)	12.3% (0-34.6)
Median LP55 (range)	1.50 cm (0-15.0)	5.40 cm (0-15.0)	2.55 cm (0-13.2)
Median LP60 (range)	0 cm (0-14.4)	1.20 cm (0-14.4)	0 cm (0-11.1)
Median LF60 (range)	0 cm (0-11.1)	0 cm (0-11.1)	0 cm (0-1.50)

Abbreviations: CTCAE = National Cancer Institute Common Terminology Criteria for Adverse Events; L_F60 = length of esophagus with full (≥90%) of circumference exposed to ≥60 Gy; L_P55 or L_{P60 =} length of the esophagus with partial (≥50%) of circumference exposed to ≥55 or ≥60 Gy, respectively; V₆₀ = percent volume of esophagus exposed to ≥60 Gy.

DLH values and prediction of dysphagia outcomes

Within the cohort of 107 patients, 6 patients developed grade ≥2 CTCAE dysphagia, and 17 developed grade ≥2 PRO-CTCAE dysphagia. CTCAE grade 1, 2, and ≥3 dysphagia occurred in 53 (49.5%), 6 (5.6%), and 0 patients, respectively. PRO-CTCAE grade 1, 2, ≥3 dysphagia occurred in 42 (32.3%), 11 (10.3%), and 6 (5.6%) patients. The proportion of grade ≥2 CTCAE dysphagia and grade ≥2 PRO-CTCAE dysphagia in the cohort was different (P = .0015, Table E1).

The esophageal partial-circumferential irradiation parameter was associated with increased odds of PRO-CTCAE ≥2 dysphagia at threshold doses of 55 Gy (L_P55) and 60 Gy (L_P60). Each 0.5 cm increase in L_P55 and L_P60 was associated with an 8% (OR, 1.08; P = .035) and a 9% (OR, 1.09; P = .029) increased odds of grade ≥2 PRO-CTCAE dysphagia, respectively (Fig. 1). The full-circumferential irradiation parameter was associated with increased odds of PRO-CTCAE grade ≥2 dysphagia when the threshold dose was 60 Gy (L_F60). Each 0.5 cm increase in the L_F60 was associated with an 11% increase in the odds of PRO-CTCAE grade ≥2 dysphagia (OR, 1.11; P = .05). The esophageal volumetric parameter V₆₀ was associated with increased odds of patient-reported grade ≥2 dysphagia. Each volumetric 2% increase in V₆₀ was associated with a 10% increase in PRO-CTCAE grade ≥2 dysphagia (OR, 1.10; P = .033). Subgroup analyses for patients with NSCLC receiving concurrent chemotherapy were consistent with the overall cohort.

Figure 1.

Esophageal full- and partial-circumferential dose-length parameters and volumetric dose parameters and log odds of developing patient-reported (PRO-CTCAE) grade ≥2 dysphagia. The logistic regression curves show the relationship between increasing esophageal length receiving partial- or full-circumferential dose exposure (L_Pdose or L_Fdose) and the odds of developing patient-reported grade ≥2 dysphagia. The odds ratios per 0.5 cm increments of L_P55, L_P60, and L_F60 demonstrate a statistically significant increase in the probability of patient-reported dysphagia, indicating that greater lengths of partial- and full-circumferential esophageal irradiation at high doses are associated with higher odds of patients reporting grade ≥2 dysphagia. The shaded regions represent the 95% CIs.

Abbreviation: CTCAE = National Cancer Institute Common Terminology Criteria for Adverse Events; Gr2 = grade 2; PRO = patient reported.

Figure 2 plots patients with various lengths of exposure to the statistically important dose-length parameters (L_P55, L_P60, and L_F60) and volumetric V₆₀ parameters, stratified by grade $\le$1 versus ≥2 dysphagia. No DVH or DLH parameter was predictive for grade ≥2 clinician-reported dysphagia. L_P55 and L_P60 demonstrated a strong correlation with V₆₀ (ρ = 0.751 and ρ = 0.755, respectively; Fig. 3). L_F60 showed a moderate correlation with V₆₀ (ρ = 0.569).

Figure 2.

Distribution of length of esophageal partial- (L_Pdose) or full-circumferential (L_Fdose) dose exposure and standard esophageal percent-volume (V_dose) exposure metrics according to clinician-reported (Part A) and patient-reported dysphagia events (Part B), stratified by grade ≤1 versus ≥2. The violin plots show that cases in which clinicians and patients report grade ≥2 dysphagia (blue) generally have greater median esophageal lengths receiving partial- or full-circumferential dose exposure to threshold doses of 50 to 60 Gy (L_P50-60, L_F60) and esophageal percent volume receiving threshold dose 60 Gy (V₆₀) compared with those with grade ≤1 dysphagia (orange). Points within each plot represent the median value, with inner box plots within each violin displaying the interquartile range at the threshold dose levels.

Figure 3.

Strength and direction of associations between length of partial- and full-circumferential esophageal irradiation at threshold doses 55 to 60 Gy (L_P55-60 and L_F60) and standard esophageal dose-volume (V₆₀) parameters. The scatter plots demonstrate the correlation between esophageal V₆₀ (percent volume of the esophagus receiving at least 60 Gy) and the length of partial-circumferential esophageal irradiation to at least 55 to 60 Gy as well as the length of full-circumferential esophageal irradiation to at least 60 Gy (L_P55, L_P60, and L_F60), as indicated by the Spearman correlation coefficients. A strong positive correlation is observed between V₆₀ and both partial-circumferential dose-length metrics, L_P55 (ρ = 0.751) and L_P60 (ρ = 0.729). The correlation between V₆₀ and full-circumferential esophageal dose-length metric, L_F60 is moderate (ρ = 0.569), indicating a similar but less pronounced association. The shaded regions represent 95% CIs for the linear fit of the data.

The likelihood ratio tests identified several DLH and DVH parameters that effectively modeled patient-reported dysphagia when compared with null models (Table 3). Specifically, higher-dose—near 60 Gy—partial-circumferential dose-length and volumetric dose metrics emerged as important, clinically relevant models with higher test statistics: L_P55 to L_P60 (4.35-4.56) and V₅₅ to V₆₀ (4.64-4.46), respectively. L_F60 evaluating the length of full-circumferential esophageal exposure to a high dose of 60 Gy was not as effective in detecting grade ≥2 PRO-dysphagia (L_F60 test-statistic: 3.61). Low-moderate dose-length and volumetric parameters also had near-high test statistics (L_F15 = 4.37; V₃₅ = 4.18), but are not clinically used esophageal dose metrics.

Table 3.

Comparison of L_Pdose, L_Fdose, and V_Dose models for predicting patient-reported dysphagia

	LPdose	LFdose	Vdose
5 Gy	1.1687	0.8549	0.3756
10 Gy	1.7705	3.588	0.5139
15 Gy	2.897	4.3745	2.0674
20 Gy	2.884	2.5933	3.6909
25 Gy	1.5884	0.9928	3.8507
30 Gy	1.3956	1.3498	4.0471
35 Gy	1.4904	1.9186	4.1797
40 Gy	2.0667	2.1648	3.9522
45 Gy	2.3908	2.2567	3.642
50 Gy	2.7924	1.7427	3.2643
55 Gy	4.3524	3.3951	4.6356
60 Gy	4.5613	3.6106	4.4585

Abbreviations: L_Pdose = length of esophagus with partial (≥50% of) circumference exposed to the threshold dose; L_Fdose = length of esophagus with full (≥90% of) circumference exposed to the threshold dose; V_dose = percent volume of esophagus exposed to the threshold dose.

Discussion

We establish esophageal partial- (L_P55, L_P60), and potentially full-circumferential (L_F60) dose-length exposure metrics as effective predictors of patient-reported esophageal toxicity. These DLH metrics employ a standardizable method of quantifying high-dose irradiation (55-60 Gy) to the partial- and full-circumference of the esophagus. Our data fill a crucial gap in the understanding of an objective measure to assess for ipsilateral dose confinement, proxy for contralateral esophagus wall sparing, and predict patient-reported dysphagia. Our data also underscore the importance of patient-reported outcome measures in evaluating dysphagia within modern-era RT.

IMRT, introduced in the early 2000s, provides more precise and conformal dose distributions compared with traditional 3-dimensional conformal RT. Although this technique was expected to reduce toxicity, a retrospective study of 165 patients with NSCLC treated with definitive chemoRT using IMRT between 2005 and 2006 found that 70% developed grade 2 esophagitis and 18% experienced grade 3 esophagitis—outcomes that were not significantly different from those observed with 3-dimensional conformal RT.^8,9 The CEST was introduced to exploit the technical capabilities of IMRT and to mitigate esophagitis severity in a phase 1 trial.⁵ Twenty-seven patients with locally advanced NSCLC or limited-stage small cell lung cancer were treated to 70 Gy, with constraints for the contralateral esophagus including D_max < 60 Gy, V₅₅ < 0.5 cm³, and V₄₅< 2.5 cm³.¹⁰ Among 25 patients, despite a median V₅₅ of 1.4 cm³ (range, 0-5.3 cm³) and V₄₅ of 2.7cm³ (range, 0-9.2 cm³), there were only 5 instances of grade 2 esophagitis and no grade 3 events. This study confirmed the feasibility of using CEST to avoid higher-grade toxicity; however, the technique allows for adjustments to substructure definitions depending on tumor proximity to the esophagus, resulting in heterogeneous contours. A retrospective study of 183 patients with stage III NSCLC receiving definitive chemoRT with IMRT showed that the use of CEST did not increase the risk of local failure nor negatively impact progression-free survival.¹¹ Our study is a proof-of-concept study that a metric that quantifies the extent of partial-circumferential esophageal irradiation can accurately assess dysphagia risk and establish appropriate dose metrics in the era of contralateral-esophageal–sparing RT.

Avoidance of circumferential organ irradiation to reduce RT toxicity has been applied to other luminal organs. In prostate cancer, IMRT and rectal balloons are used to reduce the dose to the posterior and lateral rectal wall, thereby mitigating the severity of proctitis.¹² In anal cancer, the anterior vaginal wall was contoured as an organ-at-risk and correlated with dyspareunia at a suggested optimal cutoff of V₃₅ $<$ 60%.¹³ Additionally, a circumferential vaginal canal parameter calculated the length of the vagina receiving ≥50 Gy to ≥50% of its circumference and was predictive of dyspareunia. These studies support the use of techniques designed to avoid circumferential irradiation of luminal structures and underscore the utility of our DLH parameters in guiding treatment planning and predicting toxicity in luminal organs other than the esophagus.¹³

Our study is limited by its small sample size and retrospective analysis. We report lower grade ≥2 CTCAE esophagitis rates than found in prior published studies from earlier treatment time periods.³ Our lower grade ≥2 CTCAE dysphagia may be due to the implementation of the contralateral-esophageal−sparing RT technique into clinical practice because patients were treated from 2019 to 2023. In addition, for patients at high risk of developing radiation-mucositis or esophagitis, physicians at our institution prescribe gabapentin, a non-narcotic nerve-acting medication, before anticipated dysphagia to reduce and delay the perception of severe pain and reduce the use of narcotics.¹⁴ We note the wider confidence interval when evaluating the association of esophageal DLH metrics with the odds of dysphagia compared with esophageal DVH metrics, but also highlight that DLH metrics can identify a subset of patients who receive high doses to the contralateral esophagus where the V₆₀ parameters are well below accepted treatment planning thresholds of V₆₀ < 17%. The spatial detail of DLH metrics quantifying circumferential irradiation to a luminal structure provides additional insight beyond what volumetric dose statistics can capture and may represent a previously unidentified group of patients at high risk for clinically significant dysphagia.

This study is unique in the use of robust prospectively collected CTCAE and PRO-CTCAE toxicity metrics weekly during on-treatment visits. Given the increased granularity of patient-reported over clinician-reported toxicity metrics in delineating symptoms, we were able to show that the esophageal DLH metrics L_P55, L_P60, and L_F60 can (1) quantify the length of partial- or full-circumferential esophagus exposure, (2) be associated with patient-reported dysphagia, and (3) strongly correlate with standard esophageal V₆₀ DVH metrics while adding additional spatial detail. Our future research aims to validate optimal cutoff values for L_P55, L_P60, and L_F60 in a larger cohort.

Disclosures

Chen Hu reports a relationship with Johnson & Johnson that includes: consulting or advisory. Chen Hu reports a relationship with D1 Medical Technology that includes: consulting or advisory. Todd McNutt reports a relationship with Oncospace Inc that includes: board membership. Claire Snyder reports a relationship with Pfizer Inc that includes: funding grants. Claire Snyder reports a relationship with Genentech that includes: funding grants. Claire Snyder reports a relationship with Shionogi Inc that includes: consulting or advisory. Claire Snyder reports a relationship with Movember Foundation that includes: consulting or advisory. Claire Snyder reports a relationship with Janssen Inc that includes: consulting or advisory. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.