Assessing Quality of Life and Symptoms in Transplantation and CAR-T Recipients: Expert Panel Recommendations from the Survivorship Special Interest Group of ASTCT

Abstract

Patient-reported outcomes (PROs) to measure quality of life (QOL) and other symptoms play an increasingly important role in clinical trials and regulatory approvals for hematopoietic cell transplantation (HCT) and chimeric antigen receptor T-cell (CAR-T) therapy. However, their adoption has been hindered by wide heterogeneity in the choice of PRO measures for clinical research, including the Functional Assessment of Cancer Therapy Bone Marrow Transplantation (FACT-BMT) and the European Organization for Research and Treatment of Cancer Quality of Life Core Questionnaire (EORTC QLQ-C30) inventories. Additionally, the potential for PRO integration into routine standard-of-care (SOC) practice for patients undergoing HCT or CAR-T therapy has not yet been realized. As part of a coordinated effort by three American Society for Transplantation and Cellular Therapy (ASTCT) Special Interest Groups, we developed best practices for PRO integration in adult and pediatric recipients of HCT and CAR-T therapy. We strongly encourage the use of Patient-Reported Outcomes Measurement Information System (PROMIS) or PRO version of Common Terminology Criteria for Adverse Events (PRO-CTCAE) instruments as the primary PRO measures for most HCT and CAR-T trials. Measures such as the PROMIS-29 inventory can be used for QOL assessments, while PRO-CTCAE item banks can be used for specific symptoms. Rationales for our strong recommendation to move from FACT-BMT and EORTC QLQ-C30 to PROMIS/PRO-CTCAE instruments include: (1) free licensing and ease of implementation, including in electronic medical records; (2) psychometric validation in a variety of oncologic settings, including during inpatient hospitalizations; (3) translation into multiple languages, with validation in both adult and pediatric settings; and (4) adoption into centrally-collected PRO protocols from the Center for International Blood and Marrow Transplant Research for both HCT and CAR-T recipients. Steps to operationalize these PRO measures are discussed, as are methods to migrate existing data from legacy PRO instruments. We similarly recommend the consideration of PRO integration into SOC clinical practice, including the development of threshold-based workflows to both personalize and standardize care in this setting. Other panel recommendations include the use of standardized timepoints for longitudinal PRO assessments and the inclusion of patient advocates when implementing PRO measures. Implementing these steps will improve the ability of PROs to improve outcomes for patients undergoing HCT or CAR-T therapy, both in trials and – more importantly – in SOC practice.

Introduction

Novel approaches for reducing toxicity and complications have greatly improved outcomes in patients with hematologic malignancies undergoing transplantation or cellular therapy (TCT), including hematopoietic cell transplantation (HCT) and chimeric antigen receptor T-cell (CAR-T) therapy. As these innovations continue, so too must the science of capturing the patient experience during TCT through patient-reported outcome measures (PROMs) that assess patient-reported outcomes (PROs) in a reproducible and clinically relevant manner. Longitudinal PROs are uniquely important to HCT and CAR-T therapy given the dynamic phases of these therapies, which often include extended inpatient hospitalizations with acute toxicities that are then followed by prolonged survivorship periods with the potential for durable remission. Several trials have demonstrated that PROs offer prognostic importance with regard to survival and other key health-related outcomes for these populations [1–4]. As such, both the United States (US) Food and Drug Administration (FDA) and the European Medicines Agency have encouraged PRO collection in cancer research. Several groups have issued best practices for PRO reporting in clinical trials as well [5,6].

However, there remains significant room for improvement with regard to PRO collection in TCT clinical care and trials. Although the routine incorporation of PROMs into standard-of-care (SOC) settings during HCT or CAR-T therapy is feasible (see Supplemental Table S1 and Supplemental Table S2 for annotated examples), this is not consistently implemented at most US or European centers [7,8]. Furthermore, many instruments are available to assess quality of life (QOL) and symptom burden. For HCT recipients, historical QOL measures include the Functional Assessment of Cancer Therapy – Bone Marrow Transplantation (FACT-BMT), the European Organization for Research and Treatment of Cancer Quality of Life Core Questionnaire (EORTC QLQ-C30), and the RAND Short Form Health Survey (SF-36) [9,10]. With CAR-T, over 30 different instruments have been used; these include the aforementioned instruments as well as the EuroQOL 5-dimension 5-level (EQ-5D-5L) and various Patient-Reported Outcomes Measurement Information System (PROMIS) instruments [11–13]. This heterogeneity limits the TCT field’s ability to compare and aggregate study results. More importantly, this complexity makes it challenging for clinicians to choose which PROMs to incorporate into SOC practice for patients undergoing HCT or CAR-T therapy. This manuscript describes the work of an expert panel focused on developing best practices for PROs in SOC and research settings for TCT recipients. Similar expert panels were convened regarding for three other domains of the TCT field as well: physical function assessment, cognitive function assessment, and social determinants of health (SDOH) in TCT recipients.

Methods

As part of a coordinated effort by three American Society for Transplantation and Cellular Therapy (ASTCT) Special Interest Groups (SIGs) (the Aging SIG, the Biobehavioral Research SIG, and the Survivorship SIG), we convened four expert panels in February 2024 following the annual ASTCT / Center for International Blood and Marrow Transplantation Research (CIBMTR) Tandem Meeting. Panel members were invited through each SIG or through personal contacts of SIG leadership; manuscript drafts were serially reviewed by the entirety of each panel’s authorship panel over the subsequent year.

Literature was reviewed and key recommendations (summarized in Table 1 for our panel) were developed and iteratively revised by panel members; leaders of the ASTCT Aging, Survivorship, and Biobehavioral Research SIG and key panel members (RB, BKH, ADS, CJL, MEG, RVJ, VEK) used a modified Delphi method to facilitate consensus on recommendations. Consensus was defined as all members agreeing with the statement. Because these recommendations were expected to center on expert assessments of unmet needs and best practices rather than the results of large randomized trials, formal evidence categorization were deemed impractical given the generally low quality of data in this space. As such, two tiers of recommendations were developed by the key panel members of the SIGs for use by all four panels. Tier 1 recommendations were defined by the presence of best available evidence to support their routine operationalization into clinical practice. In contrast, Tier 2 recommendations were defined as aspirational: while they potentially could improve the care of patients undergoing HCT or CAR-T therapy, further research or resources would be required before their routine use could be recommended.

Table 1:

Expert panel recommendations

TIER 1 RECOMMENDATIONS (should be operationalized)
Recommendation	Rationale
Outside of specific scenarios, use PROMIS and PRO-CTCAE measures for QOL and symptom burden in TCT trials and practice.*	PROMIS and PRO-CTCAE instruments are validated, free to use, and EMR-integrable. Data from many legacy PRO instruments can be translated to PROMIS via PROsetta [41].
Measure PROs longitudinally based on purpose or study aims, but consider landmark timepoints: pre-conditioning / pre-LDC, Day +30 (CAR-T only), Day +100, Day +180, Year 1, and yearly thereafter.	These timepoints align with important clinical landmarks during TCT as well as PRO timepoints currently being used by the CIBMTR, which can serve as a reference benchmark for future investigations.
Include patient advocates (including parents of pediatric patients) in protocol and clinical workflow development.	Patient advocates can help ensure the relevance of PRO-based research and clinical care [115].
TIER 2 RECOMMENDATIONS (aspirational but should be considered)
Recommendation	Rationale
Consider creating threshold-based referral workflows based on PRO results for pertinent symptoms during HCT or CAR-T.	Structured workflows can improve patient QOL, reduce healthcare utilization, and potentially even improve survival.
Consider structured PRO instruments† (independently of conventional clinical assessments) to assess psychological symptoms.	Structured PRO instruments can detect subclinical psychological symptoms and power symptom-guided workflows.
In the pediatric setting, collect PRO data from patients along with proxy-reported data wherever possible.	Caregivers can provide important insights about patient experience, particularly neurological symptoms.

^*For adult settings, the PROMIS-29 instrument (preferably PROMIS 29+2, which contains 2 additional items regarding cognitive function) or specific PRO-CTCAE symptom assessments are recommended. For pediatric settings, appropriate pediatric / proxy PROMIS instruments or Ped-PRO-CTCAE assessments can be used. For chronic GVHD, the mLSS has been well validated as a PRO measure and is free for academic investigators to license.

^†Suggested PRO instruments include the NCCN Distress Thermometer and PHQ-9 (see Supplemental Table S4 for details).

Abbreviations: ASTCT, American Society for Transplantation and Cellular Therapy; BMT, blood and marrow transplantation; CAR-T, chimeric antigen receptor T-cell therapy; CIBMTR, Center for International Blood and Marrow Transplant Research; EMR, Electronic Medical Record; GVHD, graft-versus-host disease; HCT, hematopoietic stem cell transplantation; LDC, lymphodepleting chemotherapy; mLSS, modified Lee Symptom Score; NCCN, National Comprehensive Cancer Network; PRO, Patient-reported outcome; PRO-CTCAE; Patient-Reported Outcomes Version of the Common Terminology Criteria for Adverse Events; PROMIS, Patient-Reported Outcomes Measurement Information System; QOL, quality of life.

These recommendations for PRO measurements in the TCT field are described and contextualized in our manuscript with dedicated emphases on both global QOL and specific TCT-related symptoms (both physical and psychological). General logistical considerations around PROM operationalization are consolidated in the Quality of Life section for ease of reading. Similarly, symptom-specific PRO considerations are discussed primarily in the Physical Symptoms section. These recommendations represent guidance from our expert panel rather than a formal position statement from the ASTCT Committee on Practice Guidelines. Nevertheless, we hope that these panel recommendations ultimately guide the TCT field toward better and broader operationalization of PROs both in clinical practice and research.

Quality of Life

QOL is a multidimensional concept that includes the effects of illness and treatment on physical and psychological functioning. In the setting of TCT, several studies have demonstrated associations between QOL and graft-versus-host disease (GVHD), treatment response, and survival [1,14–17]. In one study, a single-item question about patient-reported general health status before transplantation was a strong predictor of one-year post-HCT mortality [2]. In the setting of CAR-T, several trials have demonstrated clinically meaningful improvements of QOL in patients receiving CAR-T versus standard therapies [18–21]. However, there is a marked heterogeneity of QOL-focused PROMs in the HCT and CAR-T settings [9,11]. Historically, the FACT-BMT instrument (which contains 47–50 items depending on the version) was the most common assessment used for TCT studies. Total or subdomain FACT-BMT scores have been associated with survival, GVHD, and psychological outcomes in several HCT studies [3,15,22,23]. Similarly, the 30-item EORTC QLQ-C30 and 36-item SF-36 instruments have been associated with survival outcomes in some but not all HCT studies [1,2,4,24]. The EQ-5D-5L, which includes a visual analogue scale for patients to self-estimate their general health, has also been used to estimate quality-adjusted life-years (QALYs) as components of economic analyses in this setting [25,26]. However, the requirements for licensing with these instruments can limit their potential to be used at smaller centers and in SOC settings.

PROMIS instruments have increasingly gained prominence as the recommended tools to study QOL and other symptoms in the TCT field [9,27]. PROMIS measures, which were developed by the US National Institutes of Health (NIH), are free to use, available in over 80 languages, validated in both adult and pediatric populations, and have been psychometrically validated in a variety of oncologic settings across the world [28–34]. Practical details about accessing PROMIS instruments are listed in Supplemental Table S3. PROMIS instruments measure a variety of domains encompassing physical, mental, and social health. Logistically, they are offered as short forms of 4–10 questions per domain that can be combined into fixed collections such as the PROMIS-29 to measure global QOL. Computerized adaptive testing, which dynamically selects questions based on previous answers, is also available with PROMIS instruments to reduce survey burden. All PROMIS measures use a common metric: adjusted T-scores (with mean of 50 and standard deviation of 10) using US population norms as a reference. PROMIS measures can easily be administered electronically, including through integrations with popular electronic medical record (EMR) and electronic data capture (EDC) platforms such as Epic Systems (Verona, Wisconsin, USA) and REDCap (Nashville, Tennessee, USA) respectively [35,36].

Implementing PROMIS instruments in clinical trials

As a Tier 1 recommendation (Table 1), we strongly endorse PROMIS measures such as the PROMIS Global Health inventory (10 questions) or PROMIS-29 (29 questions) as the primary measures for global QOL in clinical trials of HCT or CAR-T therapy; symptom-specific inventories are discussed in subsequent sections of this manuscript. PROMIS measures have already been used extensively in both adult and pediatric TCT populations. To aid researchers, we have summarized over 40 presented or published investigations of PROMIS and related instruments in the HCT (Supplemental Table S1) and CAR-T settings (Supplemental Table S2); we encourage the use of data from these referenced citations as benchmarks for future research.

Additionally, the CIBMTR has now implemented a centrally-collected PRO Data Collection Protocol using PROMIS measures for adult HCT and CAR-T recipients [27]. The protocol collects PROMIS-29 data at pre-conditioning (pre-lymphodepletion for CAR-T), Day +30 (for CAR-T only), Day +100, Day +180, Year 1, and annually thereafter. While no specific windows are defined are defined in the CIBMTR PRO protocol, suggested windows are ± 7 days for Day +30, ± 14 days for Day +100, and ±28 days for subsequent timepoints. As of this publication, this protocol has collected data from over 1300 HCT recipients (both autologous and allogeneic) and over 300 CAR-T recipients. These timepoints align with important clinical landmarks during a patient’s TCT journey, and we thus recommend the use of at least these standardized timepoints for longitudinal PRO collection (a Tier 1 recommendation in Table 1). Additional timepoints may of course be necessary to align with the goals of a given study: for randomized trials involving supportive care interventions, for example, an additional PRO timepoint at symptom nadir may be clinically meaningful to collect.

As a specific example of what output might be expected from PROMIS instruments, the PROMIS-29 instrument includes eight domains: physical function, anxiety, depression, fatigue, sleep disturbance, ability to participate in social roles of activities, pain interference, and pain intensity. Many centers now specifically use PROMIS 29+2, which includes an additional 2 questions about cognitive function; we suggest the use of this specific PROMIS instrument as well. As has been done previously in the SOC setting for HCT and CAR-T recipients [37], each of these domains is scored separately using adjusted T-scores relative to the US population. Higher and lower PROMIS scores must be interpreted in the context of the domain: for example, higher anxiety T-scores are worse whereas higher physical function T-scores are better. These scores can also be summarized into a single PROMIS Preference (PROPr) summary score that ranges from approximately 0 (as bad as dead) to 1 (perfect or ideal heath) [38]. Compared to the analogous EORTC QLU-C10D summary score which is derived from EORTC QLQ-C30 responses, the PROMIS-derived PROPr score is not as affected by ceiling effects and thus – compared to EORTC QLQ-C30 data – may be more sensitive to smaller magnitudes of improvement over time [39].

Our recommendation to shift universally to PROMIS instruments in HCT and CAR-T trials is based on all of the factors above: feasibility for smaller trials and at smaller centers, in part due to free licensing; psychometric validation across a variety of oncologic settings; and adoption by organizations including CIBMTR, ASTCT, and the Blood and Marrow Transplant Clinical Trials Network (BMT CTN) [9,40]. However, we acknowledge that other PROMs have historically been used in this setting. In many such scenarios, the NIH-funded PROsetta initiative (http://prosettastone.org, details in Supplemental Table S3) may be used to calculate equivalent PROMIS scores from other instruments. The PROsetta website includes freely available conversion tables which are also available as a dedicated R package [41]. The PROsetta initiative has been validated in the TCT setting, most notably in a study of over 1600 HCT recipients where extrapolated PROMIS scores (calculated from SF-36 responses using PROsetta linkages) were highly correlated with actual PROMIS-29 data for the same timepoint [42]. In the setting of CAR-T therapy, PROsetta linkages have also been used in several studies to allow authors to modernize data from older PROMs that their institutions had previously been using for SOC data collection [43–45]. As a limitation, while crosswalks between EORTC QLQ-C30 data and PROMIS short form data have recently been published [46], PROsetta linkages do not yet exist (as of publication) for this purpose.

We acknowledge limitations to the use of PROMIS instruments, most critically the need for more granular definitions of minimal clinically important difference (MCID) thresholds during HCT and CAR-T therapy. It is important to note that, regardless of PROM, there is no single threshold will represent clinically meaningful change for all patients given heterogeneity in disease biology, therapies, timepoints, and the severity of what is being measured. While previous studies have sought to identify MCID thresholds for PROMIS instruments in patients with cancer [29,47,48], these efforts are not as mature as corresponding efforts to identify MCID thresholds for older instruments such as the EORTC QLQ-C30. That being said, a systematic review of 31 different PROMIS-based studies has concluded that an MCID of 2–6 T-score points (both between groups and for within-group comparisons) is generally adequate for PROMIS instruments [49]. Alternatively, it is reasonable to posit that half of a standard deviation reflects meaningful change for most PRO instruments. The corresponding PROMIS metric (i.e., a 5-point threshold on adjusted T-scores) has thus been used as an MCID threshold for many HCT and CAR-T studies, including several studies using the aforementioned CIBMTR PRO registry data [27,42,50–53].

Implementing PROMIS instruments in SOC settings

While PRO standardization in clinical research is important for the field, PRO implementation into SOC practice is even more important for our patients. PRO-based clinical workflows in routine practice have been shown to lower symptom burden, enhance patient-provider communication, and improve survival in several oncologic trials [54–56]. Additionally, from a scientific perspective, PRO distributions for ‘real-world’ patients may also be different than those for patients who participate in clinical trials at a TCT center [57,58]. Given the specialized infrastructure required by TCT centers to provide care to HCT and CAR-T recipients, these centers may already have much of the foundational support needed to integrate PROs into existing workflows for SOC practice. Electronic PROM delivery, either through email/web-based applications (e.g., REDCap) or directly via the EMR, has been shown to be feasible with regard to integrating PROs into standard TCT care [37]. Importantly for this population, the feasibility of electronic delivery of PROMIS and PRO-CTCAE instruments (PRO companion to Common Terminology Criteria for Adverse Events; see next section for details) has been demonstrated in multiple studies even for patients who are actively being hospitalized [52,53,59,60].

We acknowledge the competing demands that make it difficult to allocate sufficient time, staffing, technology, and other resources to administer PROMs during HCT or CAR-T. Fortunately, several resources are available for clinical practices seeking to implement PROs in routine care. HealthMeasures (https://www.healthmeasures.net/), originally funded by the NIH, is a dissemination hub for PROMIS instruments and offers detailed recommendations for PROMIS implementation into clinical practice. For centers adopting PROs for the first time, the website describes strategies to address core challenges such as defining specific aims, obtaining stakeholder buy-in, and training personnel on new workflows. Detailed guides are also available to mitigate other practical barriers with PRO implementation, for example how to stagger patient-facing requests for PROM collection from provider appointment times or how to efficiently display PROM results for frontline providers. Even for centers that have already operationalized PROs into SOC settings, the HealthMeasures website contains helpful information about best practices and quality evaluations. A second valuable resource is the Patient-Reported Outcomes Tools: Engaging Users and Stakeholders (PROTEUS) Consortium (https://theproteusconsortium.org/) website. The PROTEUS Consortium (details in Supplemental Table S3) is an international organization of clinicians, researchers, patients, policy makers, and others to support the use of PROs in research and clinical care. This group has published several resources to help with the design, implementation, and management of PRO-based workflows in a variety of practice settings [61].

Specific symptoms

Physical symptoms

Physical symptoms are an important component of QOL in HCT and CAR-T recipients. In HCT, there is a high burden of clustered physical symptoms in the acute setting including fatigue, nausea, anorexia, sleep disturbances, and pain. Although CAR-T therapy may involve fewer short-term symptomatic exacerbations, patients may still experience significant fatigue and anorexia in the first 30 days following infusion. In the longer-term setting, chronic GVHD can be a substantial source of physical symptoms in survivors of allogeneic HCT. With CAR-T therapy, hematotoxicity, infections, and neurocognitive complications may also negatively impact physical functioning. Many of these symptoms are closely intertwined with psychological symptoms such as depression and anxiety (discussed in the next section). While some common symptoms such as pain and fatigue are included in the QOL instruments described above, dedicated symptom-specific PROMs can more granularly assess the frequency and severity of these symptoms. Other symptoms relevant to the TCT field, for example dysgeusia or symptoms associated with sexual dysfunction, may not reliably be captured in general QOL instruments. Finally, longitudinal symptom-specific PROMs may demonstrate changes in symptoms during acute phases of TCT therapy (e.g., nausea with conditioning or lymphodepleting therapy) that are less likely to be registered within the 7-day recall periods used in most QOL-focused instruments.

As with QOL, there are a number of PROMs available to measure physical symptoms. As detailed with citations in Supplemental Table S1 (HCT) and Supplemental Table S2 (CAR-T), several symptom-specific PROMIS instruments have been used in this setting to investigate pain, fatigue, sleep, sexual function, physical function, and other domains. Although the MD Anderson Symptom Inventory (MDASI) has been used in this setting as well [62,63], it is not free to license or use unlike PROMIS instruments. Another important PROM to highlight is the PRO-CTCAE, which adopts National Cancer Institute (NCI)-developed terminology for adverse event reporting in clinical trials. The PRO-CTCAE is a system to evaluate symptomatic toxicity in oncology (including in the pediatric setting), with 124 items covering 78 symptomatic toxicities. PRO-CTCAE items typically involve 2–3 questions per symptom to capture symptom severity, frequency, and interference with usual or daily activities. Like PROMIS instruments, PRO-CTCAE assessments are free to use, available in multiple languages, and can be implemented into common EMR systems (e.g. the eSym Module in Epic) and R statistical packages [64,65]. Crosswalk tables to convert between PRO-CTCAE and PROMIS scores in the oncologic setting have been published as well [66].

Although no mandated scoring approach for PRO-CTCAE instruments exists, there is emerging evidence for calculated measures such as Toxicity over Time (ToxT) or the Toxicity Index; these indices can integrate longitudinal PRO data into a single metric that captures number and severity of symptoms [67–70]. Both approaches have been used in the HCT and CAR-T settings previously [71–74]. Nonetheless, formal analytic approaches are not a prerequisite to the incorporation of PRO-CTCAE inventories into routine SOC care. For example, in a large study of almost 500 patients across 26 community-based oncology practices, most patients found that PRO-CTCAE questions were easy to understand and improved discussions with their physicians [75]. Similarly, most oncologists who participated in this study felt that PRO-CTCAE assessments guided the focus of their discussions with patients and improved the efficiency of their care. As such, we recommend the integration of PRO-CTCAE items as a Tier 1 recommendation (Table 1); specific domains to be queried can be chosen by investigators and practices for specific research and SOC practices. That being said, the modified Lee Symptom Scale (mLSS) has been used uniformly in chronic GVHD to characterize a multifaceted toxicity not seen in other domains of oncology [7,76–78]. Given that the mLSS is well validated, free for investigators to license, and available in multiple languages [79], we endorse its continued use for patients with chronic GVHD.

A logical next step for the TCT field is the creation of symptom monitoring workflows to preemptively respond to worsening symptoms in real time rather than merely informing the content of the next provider visit. There is evidence of efficacy for using PRO-based workflows in oncology, most notably randomized data demonstrating significant benefits to electronic symptom monitoring in the ambulatory setting: better QOL, longer time on treatment, and improved survival [55,56,75]. However, HCT and CAR-T therapies are unique given their intensiveness from time-related and logistical perspectives. Even for hospitalized patients undergoing HCT, however, a randomized trial has demonstrated that daily electronic symptom monitoring using PRO-CTCAE assessments (with real-time results reported to inpatient nurses to then address with providers) can reduce peak symptom burden following HCT [60]. For example, with 38 patients per arm, vomiting was reported in 26% of PRO with feedback patients versus 36% of usual-care patients at Day +7, 0% versus 36% at Day +10, and 13% versus 33% at Day +14, respectively. Although specific interventions to patient-reported symptoms were not described by the authors, this study highlights that PROM operationalization in real-time peri-HCT care can improve daily symptom burden.

Issues around ensuring patient accessibility, maximizing provider bandwidth, and navigating regulatory and reimbursement-related considerations all need to be addressed before such interventions can be implemented routinely during HCT or CAR-T therapy. However, given the evidence for benefit as reviewed above, we recommend that centers consider these types of strategies to improve the patient experience for TCT recipients (a Tier 2 recommendation in Table 1). Several groups are already investigating this paradigm in the setting of allogeneic HCT, using best practices in implementation science and digital health technology to create workflows that work for patients whether they are at home or in the hospital [80,81]. As noted above, PRO-CTCAE questions have already been integrated into the Epic EMR through the eSym module [64].

Psychological symptoms

Patients undergoing HCT or CAR-T therapy must navigate several complex psychological symptoms throughout their treatment and recovery: for example, depression, anxiety, post-traumatic stress disorder (PTSD), and other affective changes [40]. In many HCT studies, depression and anxiety have been convincingly associated with worsened GVHD burden and overall survival [14,16,82,83]. While short-term psychological symptoms are normal following HCT, they may still have longer-term negative consequences: for example, if they develop into PTSD or lead to polypharmacy [84–86]. Psychological symptoms and avoidant coping may occur in up to a third of patients and are highly associated with impaired QOL, both among HCT and CAR-T recipients [87,88]. Conversely, positive engagement and approach-focused coping are associated with better QOL and improved survival [45,87–89]. With CAR-T therapy in particular, ICANS may also predispose patients to long-term psychological issues as well [90]. As such, understanding how to identify and improve these psychological symptoms through TCT and recovery is essential.

Beyond PROM heterogeneity, one unique challenge in the arena of psychological symptoms during TCT is that PROs are often not even assessed at all. In one national survey of US HCT centers, only 11% of centers used a pre-HCT PROM and very few programs performed follow-up assessments thereafter [8]. While PROMIS and PRO-CTCAE measures evaluate anxiety and depression, these PROMs have primarily been used in research settings and not in SOC care. Well-validated PROMs exist to evaluate anxiety and depression as well as other psychological symptoms, for example: the Inventory of Depression and Anxiety Symptoms (IDAS), Patient Health Questionnaire (PHQ) for depression symptoms, and the Generalized Anxiety Distress (GAD) scale for anxiety symptoms. Given the high burden of psychological symptoms among patients undergoing HCT or CAR-T therapy, we thus suggest the consideration of structured PROMs to assess psychological symptoms as an additional tool beyond conventional clinical assessments (a Tier 2 recommendation in Table 1). While PROMIS and PRO-CTCAE inventories have many advantages as described above, we acknowledge that their SOC use (e.g., the PRO-CTCAE item banks for anxious or discouraged) is currently much less common than the use of corresponding PRO-CTCAE item banks for physical symptoms. While such data mature with PRO-CTCAE instruments, validated thresholds on commonly used PROMs that might trigger automated referrals to behavioral health professionals are shown in Supplemental Table S4. Some of the unique challenges to PRO-based workflows in this domain, most notably the need to establish processes to enable longitudinal psychiatric care if needed, have been discussed elsewhere [91,92].

To be clear, these PROMs are designed to detect psychological symptoms rather than comprehensively assess psychological wellbeing. Evaluations by social work or behavioral health providers before and during TCT therapy also involve cognitive testing, assessments of psychosocial risk and protective factors (e.g., social support and positive psychological well-being), and explorations of patient care preferences that extend well beyond the topics discussed in this review [87,93,94]. Similarly, while evaluations regarding psychosocial candidacy are an essential component of the pre-TCT process, elevated scores on PROMs assessing psychosocial symptoms (e.g., elevated distress) should not be used in isolation to assess candidacy for TCT therapy. Given the ability of PRO assessments to detect subclinical symptoms as well as risk-stratify patients who need more detailed psychological evaluations, however, we believe that broader PROM incorporation into SOC practice may ultimately lead to more tailored and effective clinical assessments by behavioral health professionals.

Pediatric considerations

The implementation of PROs in the pediatric setting requires several unique considerations, some of which are outside the scope of this review designed for a broad audience. Children, adolescents, and young adults (e.g., through the age of 39) span a vast range of development abilities; as such, standards for PRO collection in each age group must vary. For younger children (e.g., 5–9 years old), the ideal PROM should incorporate simpler language, abbreviated measures, fewer options, or even pictorial scales. Notably, PRO assessments of QOL have been validated in children as young as 5 years old [95]. Adolescents (e.g., 10–19 years old) are generally underrepresented in PRO research and may require in-person reminders or monetary incentivization to encourage participation; alternatively, they may be more willing to disclose responses electronically rather than face-to-face [96,97]. Topic prioritization also varies by age group. Social function, perceived appearance, and participation in school are more likely to be relevant for children and young adolescents (e.g., 10–13 years old), while more advanced concerns including sexual health and return to work may play a larger role for older adolescents aged 14 and older [98,99]. Many PROMs have been designed for both children and adolescents living with cancer: for example, pediatric PROMIS instruments, Ped-PRO-CTCAE items, and the PedsQL Transplant Module. A pediatric version of the mLSS assessment for chronic GVHD is in development as well [100]. Inpatient PRO collection from children is feasible and may reveal short-term toxicities that can guide practice improvement [101,102].

While the child’s perspective should be considered the gold standard for PRO research in the TCT space, parent or caregiver proxy data can also add important insights. Discordance between patient and proxy reports of symptom burden is predictable and can be expected to some extent, in particular for pain and anxiety that can be harder for others to objectively assess [103–105]. However, a proxy report may be the best approximation of the child’s lived experience for patients who are too young, too ill, or cognitively unable to complete a survey. Even for children who can complete PRO assessments, concurrently collecting both the patient and proxy report can provide a richer description of a patient’s symptoms or function [106]. As a TCT-specific example, a caregiver’s perception of a patient’s cognitive function during ICANS after CAR-T therapy may reveal patterns that otherwise would not emerge from patient report alone [107]. For children with neurodivergent function, this consideration may be especially relevant. We thus recommend the consideration of PRO data from both patients and proxies wherever possible (a Tier 2 recommendation in Table 1), although we recognize that standards in the collection and analysis of concurrent self- and proxy-reported data (including methods for managing reconciliation of reporter disagreement) remain areas for future investigation [106].

As an important area for future research, we also acknowledge the importance of studying the drivers of QOL in the parents and caregivers of pediatric TCT recipients. Child wellbeing is intimately associated with parent wellbeing in the setting of cancer. Multiple studies in the pediatric TCT setting have shown that children report worsened QOL when their parents experience poor QOL or high psychosocial distress [108–110]. Most alarmingly, up to a third of parental caregivers express suicidal ideation in the three months following their child’s HCT or CAR-T infusion [108]. In both the HCT and CAR-T settings, parental caregivers often note financial strain as a significant contributing factor to psychosocial distress [111,112]. While a comprehensive review of financial toxicity in the TCT setting is outside the scope of this manuscript, proactive referrals of at-risk patients to financial navigators or social workers may help to address adverse SDOH through financial counseling and disability applications.

Discussion

There is compelling evidence that PROs add value to research and clinical care across the TCT field. The importance of capturing the patient experience in trial protocols is reflected in the experience of the BMT CTN, which has incorporated PROs into almost every study protocol since its initial establishment in 2001. Upcoming BMT CTN recommendations regarding the collection and interpretation of PRO data in TCT trials will offer complementary guidance. Beyond the setting of trials, PRO-based workflows in routine practice are associated with lower symptom burden, enhanced patient-provider communication, and improved survival [54–56]. It is no surprise that PROs are a growing component of quality and regulatory metrics as well, for example with FDA drug approvals or with distress screening as an accreditation standard for US cancer centers [113,114].

At the same time, several barriers to the optimal implementation of PROs into TCT workflows exist. The heterogeneity of PROMs for QOL and symptom burden – coupled with the licensing fees inherent to many older instruments such as FACT-BMT and EORTC QLQ-C30 – has hindered their adoption at smaller centers and prevented most clinicians from understanding their nuances. Broadly speaking, there should not be one set of PROMs used for clinical TCT research and a second set of PROMs used for routine TCT care. Our position paper describes the impetus to change this status quo and summarizes many resources for practices interested in operationalizing this process. Our vision for the future (Figure 1) includes widespread PRO adoption across the HCT and CAR-T fields (both in research and SOC settings) with the use of standardized methods and workflows to mitigate barriers to PRO implementation. Despite the limitations inherent to PROMIS instruments (most notably that they were not designed specifically for TCT recipients), their free nature and ease of implementation constitute powerful arguments in favor of their preferential use in this setting. Indeed, as advancements in the HCT and CAR-T fields increasingly allow patients to lead more normal lives following cell therapy, the benchmarking of PROMIS scores to the general US population will increasingly become an asset of these instruments rather than a drawback.

Figure 1: Future vision for PROs in the HCT and CAR-T fields.

* The NIH-funded PROsetta initiative (which can convert data from legacy instruments to PROMIS data) has been used in several HCT and CAR-T studies to date; see text for details.

Abbreviations: CAR-T, chimeric antigen receptor T-cell therapy; EMR, electronic medical record; EORTC QLQ-C30, European Organization for Research and Treatment of Cancer Quality of Life Core Questionnaire; EQ-5D-5L, EuroQOL 5-dimension 5-level; FACT-BMT, Functional Assessment of Cancer Therapy – Bone Marrow Transplantation; HCT, hematopoietic stem cell transplantation; MDASI, MD Anderson Symptom Inventory; mLSS, Modified Lee Symptom Score; NIH, National Institutes of Health; PRO-CTCAE; Patient-Reported Outcomes Version of the Common Terminology Criteria for Adverse Events; PROMIS, Patient-Reported Outcomes Measurement Information System; SF-36, Rand Short Form Health Survey; SOC, standard of care.

That being said, more important than which PROM is used is whether that PROM is used well. To some extent, PROs will always be susceptible to error and non-response biases that require methodological safeguards to guarantee their validity and reproducibility – for example, the choice of MCID thresholds and the handling of missing data. PROMs can lead to survey burden and potential psychological distress from long or intrusive question inventories. As such, multiple studies have demonstrated the value of soliciting patient perspectives in developing patient-friendly research protocols [115–117]. This goal is particularly important to patient-reported outcomes, which can only capture the patient voice if they measure symptoms and concepts that are relevant to patients. The BMT CTN has previously endorsed the incorporation of patient and caregiver research partners as a step to increase the relevance of PRO-based research in the TCT field [115]. As a Tier 1 recommendation (Table 1), we recommend the inclusion of patient and caregiver research partners (including parents of pediatric patients) in protocol development and workflow implementation whenever PROMs are being considered for clinical trials. Similarly, we recommend patient advocates to be included in all endeavors to operationalize PROs in SOC practice.

In addition to what PROMs are chosen and whom they are chosen by, another important consideration is the need to ensure access for all patients. While electronic PRO assessments in the peri-TCT setting have several advantages over in-person instruments [53,59], centers must provide adequate support for vulnerable patients as well. For patients without written English proficiency, in-person or phone-based survey administration are important measures to ensure their inclusion in PRO-based research and workflows. Given that lack of Internet access is an independent predictor of worsened PROs following HCT even after adjusting for patient location and driving distances [118], such tools may be helpful for patients without personal computing devices as well. Finally, every effort should be made to use validated translations of PROMs in the native languages of patients being asked to complete them. As one example of the complexity of this process, an early Japanese version of the SF-36 instrument led to unexpected results due to patients misunderstanding the word limited (with regard to physical limitations that affected their daily functioning) to include physician-recommended limitations on physical function as well [119]. As of publication, validated translations exist in over 70 languages for the PROMIS-29 inventory (details in Supplemental Table S3) and 12 languages for the mLSS assessment [79].

This manuscript has many important limitations. Most importantly, we intentionally focused on the practical operationalization of PROMs into clinical trials and SOC practice rather than on the underlying science of instrument development, assessments of psychometric properties, and statistical validation. In reality, rigorous steps to validate the performance of the PROMs described in this manuscript will be essential as they are increasingly deployed in various TCT settings. As another limitation, we were unable to review all possible aspects of QOL after HCT or CAR-T therapy, for example the more complex components of human functioning such as social integration and return to work/school. We also acknowledge that our review of PROs in the pediatric population is limited given the broad scope of this paper; however, separate pediatric-focused PRO recommendations are currently in development through the Children’s Oncology Group and CIBMTR. In addition, although we separated the discussion of multidimensional QOL versus physical symptoms versus psychological symptoms in our review, in reality these factors are closely intertwined and may affect each other. Similarly, the impacts of SDOH on these measures are being reviewed by a separate expert panel. Finally, our recommendations are most relevant to providers practicing in high-income countries in North America and Europe. Inadequate staff support and lowered patient literacy, among other challenges, may present unique challenges to PRO operationalization for TCT providers in low- and middle-income countries [120,121].

In conclusion, we have proposed best practices for systematically collecting and analyzing PROMs after HCT and CAR-T therapy. A central theme of our recommendations is the adoption of PROMIS and PRO-CTCAE instruments as the default PROMs in both TCT trials and SOC practice. We are mindful of the fact that achieving this future state will require buy-in from patients, providers, researchers, and other stakeholders to take an active role in the development of future trial protocols and clinical workflows. Nonetheless, we are confident that implementing this approach will allow our field to leverage PROs to improve TCT treatment modalities in a patient-centered and scientifically rigorous way.

Highlights.

• Patient-reported outcomes (PROs) play an important role in HCT and CAR-T therapy

• Integrating PROs into practice for HCT/CAR-T recipients can improve care

• We describe recommendations to standardize PRO use in HCT/CAR-T trials and practice

• PROMIS and PRO-CTCAE measures should be considered for all HCT/CAR-T trials