Domains of prognostic awareness, quality of life, and psychological distress in patients with advanced cancer

Abstract

Introduction

Prior studies have shown mixed findings regarding the relationship between prognostic awareness with quality of life (QOL) and psychological distress in patients with advanced cancer. Prognostic awareness is a multidimensional construct including a cognitive component (ie, the ability to understand one’s life-limiting illness) and an emotional coping component (ie, the capacity to process terminal prognosis). Yet, it remains unclear which domains of prognostic awareness are associated with QOL and psychological distress.

Materials and Methods

We conducted a cross-sectional study of adults with metastatic solid tumors treated with noncurative intent at a single academic center from 11/2019 to 6/2022. We used the Prognostic Awareness Impact Scale (PAIS) to measure components of prognostic awareness, including the cognitive and emotional coping components. We used the Functional Assessment of Cancer Therapy-G and Hospital Anxiety and Depression Scale to measure QOL and psychological distress, respectively. Linear regression models were used to examine the relationship between the PAIS domains and patient-reported outcomes.

Results

We enrolled 61.9% (632/1021) of eligible patients. Cognitive understanding of prognosis was not associated with QOL (B = −2.3, P = .114), anxiety (B = 0.7, P = .057), or depression symptoms (B = 0.4, P = .293). However, higher emotional coping with prognosis was associated with better QOL (B = 1.7; P < .001), lower anxiety (B = −0.6; P < .001), and depression (B = −0.3; P < .001).

Conclusion

Patients’ emotional coping with their prognosis, rather than cognitive acknowledgment of their incurable cancer, was associated with QOL and psychological distress. Our findings underscore patients’ ability to tolerate an accurate understanding of their prognosis and the critical need to incorporate effective coping strategies during prognostic discussions.

Implications for Practice.

Emotional coping with prognosis, rather than cognitive acknowledgment of it, was associated with better quality of life (QOL) and lower psychological distress in patients with advanced cancer. These findings highlight the importance of incorporating effective coping strategies into prognostic discussions in advanced cancer care, guiding healthcare providers in better supporting patients to manage their emotional responses to prognosis and ultimately optimizing their well-being.

Introduction

Patients with advanced cancer often face difficult and complex decisions regarding their medical care and treatment options.^1,2 To make fully informed medical decisions following a cancer diagnosis, patients must have an understanding of the likely course of their disease. Unfortunately, many patients report an inaccurate understanding of their likely disease trajectory and prognosis.^3,4 More recent literature suggests that prognostic awareness is a multidimensional construct that incorporates^5–9 (1) patients’ cognitive understanding of their prognosis (ie, one’s ability to acknowledge their life-limiting illness), (2) emotional coping with prognosis (ie, one’s ability to cope with both an incurable prognosis as well as prognostic uncertainty, and (3) adaptive response (one’s capacity to utilize their prognostic awareness to inform their life and medical decisions).^10,11

Prognostic discussions between clinicians and patients, crucial to fostering patient-centered care throughout the care continuum, are ongoing, not a one-time encounter.¹² However, some oncology clinicians are often reluctant to engage in any prognostic discussions, given concerns about diminishing patient’s hope. In fact, oncology clinicians delivering prognostic information about poor outcomes have been perceived by patients to be less compassionate and trustworthy than those delivering optimistic messages.¹³ Additionally, prior work investigating the association between prognostic awareness and patient-reported QOL and psychological distress has yielded conflicting results and contributed to oncologists’ ambivalence toward these discussions.^4,14–17 Consequently, these findings have created uncertainty for oncology clinicians regarding whether prognostic disclosures can be potentially harmful to patients with advanced cancer.¹⁸

Despite the multidimensional nature of prognostic awareness, prior studies have not examined which domains of prognostic awareness are associated with patient-reported QOL and psychological distress. Understanding the relationship between prognostic awareness domains and patient-reported outcomes may shed light on some of the aforementioned conflicting findings and inform how to best discuss prognosis in patients with advanced cancer. Additionally, identifying patient factors that are associated with various domains of prognostic awareness can be especially helpful in identifying populations at risk for difficulty with cognitive acknowledgment of their prognosis or capacity to emotionally cope with their incurable cancer. ¹² Yet, there is no data to date examining the association between patient factors and various domains of prognostic awareness.

In this cross-sectional study in patients with advanced incurable solid cancers, we used the Prognostic Awareness Impact Scale (PAIS), a recently developed multidimensional instrument to assess the cognitive, emotional, and adaptive domains of prognostic awareness,^10,11 and examined the association of patient- and disease-related factors with these domains. We then examined the relationship between prognostic awareness domains with patient-reported QOL and psychological distress.

Materials and methods

Study design

We conducted a single-site, cross-sectional study with metastatic solid tumors at Massachusetts General Hospital in Boston, MA. The Dana-Farber/Harvard Cancer Center Institutional Review Board approved this study. We obtained informed consent from all participants. Participants had the option to complete assessments in person during a clinic visit or electronically via a secure web-link survey.

Participants

Patients were eligible if they were adults (≥18 years of age) with advanced solid tumors, defined as cancer not being treated with curative intent according to documentation in the electronic health record (EHR). They also required the ability to comprehend English and complete surveys with minimal assistance from an interpreter, as the Prognosis Awareness Impact Scale (PAIS) was only available in English. Patients were excluded if they: (1) only had one oncology visit at the MGH Cancer Center and (2) if the treating clinician believed they would not be able to participate in study procedures and provide informed consent due to a severe psychiatric disorder (eg, psychosis, bipolar disorder, major depressive disorder) or another co-morbid disease (eg, dementia, cognitive disorder).

Recruitment methods

To identify potentially eligible patients, a clinical research coordinator sequentially screened the outpatient oncology clinic schedules at the cancer center to ensure participants had diverse types of cancers. With permission from their oncologists and to minimize sampling bias, the research coordinator approached all eligible patients consecutively and enrolled them either in person during scheduled clinic visits or over the phone. Study participation entailed completing the study assessments.

Sociodemographic and clinical data

Participants self-reported sociodemographic information, including age, gender, race, ethnicity, religion, relationship status, education, and income. We obtained information on cancer type, Eastern Cooperative Oncology Group (ECOG) performance status, and type of therapy (eg, chemotherapy, targeted therapy, immunotherapy) from the EHR.

Prognostic awareness impact scale

We used the PAIS to assess the cognitive, emotional, and adaptive domains of prognostic awareness.^10,11 The PAIS development has been recently described along with its conceptualization of the multidimensional nature of prognostic awareness.^10,11 The Cronbach alpha for the PAIS is 0.841. The cognitive domain of the PAIS focuses on assessing patients’ cognitive understanding of their prognosis characterized using the following question: (1) “Did your oncologist say your cancer was curable?” (The three response options are “yes,” “no,” or “my oncologist has not said whether my cancer is curable”). The emotional domain of the PAIS captures patient’s capacity to emotionally integrate their prognosis and cope with its terminal nature as well as cope with prognostic uncertainty. The emotional coping domain comprises ten items with scores ranging from 0 to 30. Higher scores indicate better emotional coping with prognosis. The Cronbach alpha for the emotional domain is 0.794. Finally, the adaptive response domain of the PAIS captures patient’s ability to utilize prognostic information to inform their life and medical decisions. The adaptive response domain consists of 12 items with scores ranging from 0 to 36. Higher scores indicate better adaptive response. The Cronbach alpha for the adaptive response domain is 0.871. The PAIS is included as Supplementary Table 1.

Patient-reported outcome measures

Quality of life

We used the 27-item Functional Assessment of Cancer Therapy-General to assess patient reported QOL in all study participants. The FACT-G consists of four subscales: physical well-being, social well-being, emotional well-being, and functional well-being.¹⁹ Score ranges 0-108 with higher scores indicate better QOL.

Anxiety and depression symptoms

We used the Hospital Anxiety and Depression Scale (HADS) to assess symptoms of depression and anxiety in all study participants. The HADS is a 14-item questionnaire that contains two 7-item subscales assessing depression and anxiety symptoms during the past week.²⁰ The score range for each sub-scale is 0-21. Higher scores indicate worse anxiety or depression symptoms.

Statistical analysis

We performed all statistical analyses with STATA 18.0 (StataCorp, College Station, TX). We used descriptive statistics (eg, mean) to summarize patients’ baseline characteristics for continuous variables and proportions for categorical variables. Additionally, we used unadjusted analyses to evaluate the association between patient (eg, age, gender, race/ethnicity, relationship status, religion, education level) and clinical characteristics (eg, cancer type, types of therapy, ECOG performance status) with the cognitive, emotional, and adaptive response domains of PAIS. Unadjusted analyses informed the selection of patient and clinical characteristics (ie, covariates) to be adjusted for in our multivariate regression models.

To examine the relationship between the PAIS domains with patient-reported QOL, we used multivariate regression models adjusting for age, gender, race/ethnicity, marital status, cancer type, type of therapy, ECOG performance status, time from advanced cancer diagnosis to enrollment based on findings from our unadjusted analyses and because prior work has shown these factors are associated with prognostic awareness.^21–25 As we are interested in assessing the relationship between each PAIS domain with QOL, we built three separate multivariate regression models (adjusting for variables mentioned above) to examine the relationship between each PAIS domain with QOL. We leveraged an identical approach when examining the relationship between the PAIS domains with psychological distress. We assessed and did not identify substantial collinearity within each model with the Variance Inflammation Factor (VIF) < 2. With a low overall rate of missing data, we conducted complete case analyses without accounting for missingness. We considered a two-sided P-value < .05 as statistically significant for all analyses. Although these analyses are intended to be exploratory and hypothesis-generating, we used the False Discovery Methods (FDR of 0.05) to correct for multiple testing in the multi-variate regression models.

Results

Participant characteristics

We enrolled 61.9% (632/1021) of eligible patients from November 2019 to June 2022. Table 1 summarizes participant characteristics. Participants had a mean age of 66.4 (SD = 11.3), 50.3% (n = 318) were female, 98.3% (621/632) were non-Hispanic White, and 76.3% (482/632) were married or living with someone. The four most common cancer diagnoses were genitourinary (18.7%; 118/632), lung (17.1%; 108/632), breast (16.5%; 104/632), and gastrointestinal (16.1%; 102/632).

Table 1.

Participant characteristics.

	Total
	N = 632
Age	66.4 (11.3)
Sex
Male	314 (49.7%)
Female	318 (50.3%)
Ethnicity
Hispanic/Latino	10 (1.6%)
Non-Hispanic	621 (98.3%)
Missing	1 (0.2%)
Race
American Indian/Native	7 (1.1%)
Asian	21 (3.3%)
African American	12 (1.9%)
White	581 (91.9%)
Other	11 (1.7%)
Relationship status
Married or Living with someone	482 (76.3%)
Non-cohabiting relationship	12 (1.9%)
Single, never married	45 (7.1%)
Divorced/Separated	46 (7.3%)
Widowed/Loss of long-term partner	41 (6.5%)
Missing	6 (0.9%)
Religion
Catholic Christian	268 (42.4%)
Other Christian	161 (25.5%)
Jewish	43 (6.8%)
Atheist	20 (3.2%)
None	95 (15.0%)
Other	36 (5.7%)
Missing	9 (1.4%)
Education
High school diploma (GED) or less	109 (17.2%)
Some college/college degree	296 (46.8%)
Post-graduate, professional, or doctorate degree	224 (35.4%)
Missing	3 (0.5%)
Income
<$50 000	128 (20.2%)
$50 000-$100 000	139 (22.0%)
>$100 000	290 (45.9%)
Missing	75 (11.9%)
Cancer type
Lung	108 (17.1%)
Head, neck and thyroid	42 (6.6%)
Gastrointestinal	102 (16.1%)
Melanoma	41 (6.5%)
Sarcoma	30 (4.7%)
Genitourinary	118 (18.7%)
Gynecological	35 (5.5%)
Brain	52 (8.2%)
Breast	104 (16.5%)
Line of therapy
chemotherapy	212 (33.5%)
Targeted therapy	281 (44.5%)
Immunotherapy	139 (21.9%)
Time from advanced cancer dx, median (range)	271 (1-5371)
ECOG
0	242 (38.3%)
1	323 (51.1%)
2	54 (8.5%)
3	12 (1.9%)
4	1 (0.2%)

Patient and clinical factors associated with PAIS domains

Table 2 summarizes the relationship between patient factors and prognostic awareness domains. Overall, 80.7% (502/622) of patients reported that their oncologist stated their cancer was not curable. Age, gender, being non-Hispanic white, marital status, religion, education, cancer type, months from diagnosis of advanced cancer to enrollment, and type of therapy were not associated with patient’s cognitive understanding of their prognosis. Higher patient ECOG performance status was associated with higher odds of having an accurate cognitive understanding of prognosis (OR = 1.3, P = .056), but this association was not statistically significant.

Table 2.

Unadjusted association between cognitive PAIS domain, emotional PAIS domain, and adaptive PAIS domain, sociodemographic and clinical factors.^a

	Cognitive PAIS domain		Emotional PAIS domain		Adaptive PAIS domain
Variable	OR	P-value	B	P-value	B	P-value
Age at enrollment	1.0	.806	0.05	.002	−0.05	.033
Cancer type
Lung	Ref		Ref		Ref
Gastrointestinal	1.4	.381	−0.2	.719	−0.3	.739
Breast	1.2	.689	−0.9	.145	−1.3	.122
GU	0.8	.530	0.2	.745	−1.8	.028
Other	0.9	.938	−1.1	.042	−1.1	.114
Gender
Women	1.4	.105	−1.4	<.001	0.2	.647
Non-Hispanic White
Yes	1.5	.215	−0.2	.671	−2.3	.005
Relationship status
Married or Living with someone	0.8	.382	0.6	.168	0.6	.345
Religion
Catholic Christian	Ref		Ref		Ref
Protestant	0.9	.604	0.1	.769	−0.7	.232
Other	1.2	.518	0.4	.297	−0.9	.123
Education level
High school graduate/GED	Ref		Ref		Ref
Some college/College degree	1.1	.712	0.1	.887	−0.8	.268
Post graduate	1.2	.502	0.9	.082	−1.3	.079
ECOG performance status	1.3	.056	−0.4	.144	−0.3	.437
Types of therapy
Chemotherapy	Ref		Ref		Ref
Targeted therapy	1.4	.116	−0.1	.774	−0.9	.114
Immunotherapy	0.9	.810	0.1	.779	−0.1	.939
Months from diagnosis of advanced cancer	0.99	.923	0.02	<.001	0.01	.367

^aOR = Odds ratio used for cognitive PAIS domain; OR > 1 indicate higher odds of accurate cognitive understanding of prognosis. B = unstandardized coefficient examining the association between sociodemographic factors with the emotional and adaptive PAIS domains. Bold values indicate statistically significant results.

The mean score of the emotional domain of the PAIS was 19.2 (SD: 4.1). While older age was associated with higher emotional coping with prognosis, female gender was associated with lower emotional coping with prognosis. The mean score of the adaptive response domain of the PAIS was 22.9 (SD: 5.9). Older age, having genitourinary cancer, and being non-Hispanic White were associated with a lower adaptive response to prognosis.

Association between PAIS domains and patient-reported QOL

Table 3 summarizes the adjusted analysis of the association between the PAIS domains with patient-reported QOL. After controlling for patient and disease characteristics, patients’ cognitive understanding of their prognosis was not associated with their QOL. In contrast, higher scores on the emotional coping with prognosis domain (B = 1.7, P < .001) and adaptive response domain (B = 0.3, P = .004) were associated with better QOL. These findings remained significant after correcting for multiple testing using the FDR method.

Table 3.

Multivariate association between prognostic awareness domains and patient-reported quality of life.^a

QOL	B coefficient	P-value	95% confidence interval
Cognitive understanding of prognosis	−2.3	.114	−5.2	0.6
Emotional coping with prognosis	1.7	<.001	1.4	2.1
Adaptive response to prognosis	0.3	.004	0.1	0.5

All models controlled for age, gender, cancer type, race/ethnicity, performance status, type of therapy, marital status, time from advanced cancer diagnosis to enrollment.

^a B = unstandardized linear regression coefficient describing the effect on QOL for every one-point increase in prognostic awareness domains. Bold values indicate statistically significant results.

Association between PAIS domains and patient-reported psychological distress

Table 4 summarizes the adjusted analysis of the association between the PAIS domains and anxiety symptoms. After controlling for patient and disease characteristics, patients’ cognitive understanding of their prognosis was not associated with their anxiety symptoms. However, higher scores on the emotional coping with prognosis domain was associated with lower anxiety symptoms (B = −0.6, P < .001). Higher scores on the adaptive response domain were associated with higher anxiety symptoms (B = 0.1, P < .001).

Table 4.

Multivariate association between prognostic awareness domains and patient-reported anxiety symptoms.^a

Anxiety symptoms	B coefficient	P-value	95% confidence interval
Cognitive understanding of prognosis	0.7	.057	−0.02	1.3
Emotional coping with prognosis	−0.6	<.001	−0.6	−0.5
Adaptive response to prognosis	0.1	<.001	0.03	0.1

All models controlled for age, gender, cancer type, race/ethnicity, performance status, type of therapy, marital status, time from advanced cancer diagnosis to enrollment.

^a B = unstandardized linear regression coefficient describing the effect on anxiety symptoms for every one-point increase in prognostic awareness domains. Bold values indicate statistically significant results.

When it comes to depression symptoms, in multivariate analyses, patients’ cognitive understanding of their prognosis was not associated with depression symptoms. Higher scores on emotional coping with prognosis domain (B = −0.3, P < .001) and adaptive response (−0.05, P = 0.040) were associated with lower depression symptoms (Table 5). These findings remained significant after correcting for multiple testing using the FDR method.

Table 5.

Multivariate association between prognostic awareness domains and patient-reported depression symptoms.^a

Depression symptoms	B coefficient	P-value	95% confidence interval
Cognitive understanding of prognosis	0.4	.293	−0.3	1.1
Emotional coping with prognosis	−0.3	<.001	−0.4	−0.2
Adaptive response to prognosis	−0.05	.040	−0.1	−0.01

All models controlled for age, gender, cancer type, race/ethnicity, performance status, type of therapy, marital status, time from advanced cancer diagnosis to enrollment.

^a B = unstandardized linear regression coefficient describing the effect on anxiety symptoms for every one-point increase in prognostic awareness domains. Bold values indicate statistically significant results.

Discussion

Our findings indicate that cognitive understanding of prognosis is not associated with QOL or psychological distress in patients with advanced cancer. Instead, emotional coping and adaptive response to prognosis are associated with these outcomes. These results have important clinical implications for prognostic disclosure, highlighting the multidimensional nature of prognostic awareness and the need for a nuanced examination of its impact on patient-reported outcomes and care, rather than treating it as a single construct.^26–28

Contrary to prior work,^16,29 we show that patients with an accurate cognitive understanding of their prognosis do not necessarily experience worse QOL or psychological distress. These data highlight that prognostic disclosure and the cultivation of an accurate cognitive understanding alone may not cause harm to patients. Specifically, our findings suggest that psychological distress symptoms, such as anxiety and depression, which may accompany prognostic discussions, may not solely arise from a patient’s cognitive understanding of negative or uncertain news about their illness.^16,18 Instead, factors related to cognitive understanding,³⁰ such as emotional and adaptive resources for ­managing information about their illness and trajectory, are more likely to influence distress and other negative reactions to prognostic disclosure.³¹ Thus, while effective communication of prognostic information is crucial,^32,33 oncology clinicians should also be equipped to facilitate prognostic discussions that extend beyond merely conveying facts about an illness and recovery trajectory to a heightened awareness of the relevance of resiliency and coping skills training for patients during these discussions.

Our observation that patients with higher emotional coping with prognosis and better adaptive response have lower psychological distress and better QOL may help explain inconsistencies in previous findings about prognostic awareness.^26–28 Emotional coping likely facilitates the emotional integration of prognostic information and moderates the impact of prognostic awareness on QOL and psychological well-being.^16,34 Importantly, prognostic disclosure that solely focuses on the facts of terminal illness (ie, the cognitive aspects of prognosis) without addressing emotional coping strategies may be insufficient and potentially detrimental to QOL and psychological well-being. Given that emotional coping evolves over time,³⁵ our findings suggest that a nuanced understanding of how patients process and cope with their terminal illness is imperative to further tailoring prognostic discussions for patients with advanced cancer. These new insights call for future prospective longitudinal and causal inference studies to examine how prognostic awareness changes over time and to explore the potential bidirectional associations between prognostic awareness domains and patient-reported outcomes, including psychological distress and QOL. This research could inform practical and timely supportive interventions aimed at facilitating effective coping and adaptive responses as patients navigate terminal prognoses.

We show that sociodemographic (eg, age, gender, ethnicity) and clinical factors (eg, line of therapy, type of cancer ­diagnosis) are associated with certain domains of prognostic awareness but not others. For example, older age and male gender were associated with better emotional and adaptive domains of prognosis, but not with the cognitive domain. Our findings corroborate prior work indicating that older adults (ie, ≥65-year old) have lower prognostic understanding of their disease curability and life expectancy estimates than younger adults. This may be due to a higher risk of potential cognitive impairments with aging, which could explain the observed lower cognitive understanding of prognosis.^36–39 On the contrary, increased emotional regulation observed in older adults may help explain the higher emotional and adaptive responses to prognosis we observed.^40,41 Additionally, prior studies suggest that males may be more likely to use problem-focused coping strategies when faced with distressing information, which could contribute to greater emotional and adaptive responses in the context of prognostic information.⁴² Interestingly, we found no association between race/ethnicity or age and the cognitive domain of prognostic awareness. This contradicts commonly held beliefs that older individuals have poorer cognitive prognostic awareness, as well as prior literature showing that Black race is associated with inaccurate perceptions of prognosis.^21–25 One reason for this discrepancy may be the difference in how prognostic understanding is conceptualized; while prior studies have examined prognostic awareness as a single construct, our study distinguishes between cognitive, emotional, and adaptive domains. Additionally, patients who identified as non-Hispanic White reported lower scores in the adaptive response domain, suggesting a potential lower capacity to use prognostic information in planning for the future—a finding that may reflect cultural differences in attitudes toward medical and life decisions or coping ­strategies.^43–45 These findings underscore the complexity of prognostic awareness and highlight the need to examine sociodemographic factors in the context of specific prognostic awareness domains. Sociocultural influences (eg, culture) likely moderate how these characteristics shape patients’ understanding and response to prognosis.⁴⁶

This study has noteworthy limitations. First, although our study sample was large and included patients with diverse cancers, participants were recruited from a tertiary care academic center, and most participants being White, non-Hispanic, with higher levels of education. As such, our depiction of prognostic awareness based on the PAIS may not be directly applicable to non-English-speaking individuals. Future research is essential to assesses prognostic awareness using validated measures specifically designed for non-English speaking patients. Second, although the PAIS was rigorously developed through expert consensus and cognitive interviews with an English-speaking population, and the cognitive domains have now been shown to be responsive to change in palliative care intervention ­trials,⁴⁷ it is still relatively new and has not yet undergone formal psychometric validation. Furthermore, the conceptualization of prognostic awareness as a multidimensional construct has been previously published,^10,12,28 but is not agreed upon in the field.^30,48 Additionally, questions assessing patients’ cognitive understanding of prognosis relied heavily on their perception of their prognosis based on engaging in discussions with their clinicians. This strategy to assess cognitive understanding of prognosis to avoid measuring patients’ hopes for a cure rather than their realistic cognitive understanding of their prognosis has been reported in the literature, but nonetheless has its own limitations.^10,28 Third, due to the dynamic nature of prognostic awareness, we were not able to account for changes in prognostic awareness and its impact on outcomes over time. Fourth, our cross-sectional study design also did not allow us to examine the potential bidirectional relationship between patient-reported outcomes, such as psychological distress, and PAIS domains like adaptive response to prognosis. Lastly, we did not correct for multiple testing in this study and findings should be considered hypothesis-generating.

In conclusion, this study demonstrates that patients’ emotional coping and adaptive response to prognosis, rather than their cognitive acknowledgment of their incurable cancer, were associated with their QOL and psychological distress. Therefore, further research is necessary to unravel the inherent complexity and multidimensional nature of prognostic awareness and its associations to patient outcomes, offering valuable insights for clinical practice.

Author contributions

Hermioni L. Amonoo (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing), Joely Centracchio (Writing—review & editing), Claire ­Greydanus (Data curation, Investigation, Writing—review & editing), Mitchell W. Lavoie (Writing—review & editing), Emma P. Keane (Writing—review & editing), Joseph A. Greer (Writing—review & editing), ­Elizabeth C. Lindenberger (Writing—review & editing), Keri Brenner (Writing—review & editing), Nneka N. Ufere (Writing—review & editing), Lara Traeger (Writing—review & editing), Jennifer S. Temel (Writing—review & editing), Elyse R. Park (Writing—review & editing), and Vicki A. ­Jackson (Writing—review & editing), Areej El-Jawahri (Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Validation, Visualization, Writing—original draft, Writing—review & editing)

Supplementary material

Supplementary material is available at The Oncologist online.

Funding

National Cancer Institute through grant K08CA251654 (to Dr. Amonoo) and the Doris Duke Foundation Clinician-Scientist Development Award (to Dr. Amonoo).

Conflicts of interest

A.E.J. is a consultant for Incyte Corporation, GSK, Novartis, and Tuesday Health. V.J. is a consultant for Tuesday Health. J.A.G. serves as a consultant for BeiGene, is on the speaker bureau for GSK, receives research funding from Blue Note Therapeutics, and earns royalties from Oxford University Press. All other authors have no conflicts to report.

Data availability

For original data, please contact hermioni_amonoo@dfci.harvard.edu.