Longitudinal assessment of quality of life in indolent non-Hodgkin lymphomas managed with active surveillance

Abstract

There are limited data describing the impact of active surveillance on longitudinal health-related quality of life (HRQoL) in patients with indolent non-Hodgkin lymphomas (NHL). A cohort of untreated indolent NHL patients completed FACT-LYM questionnaires at 6, 12, 18, 24, and 36 months after diagnosis. Longitudinal FACT-LYM scores were analyzed by ANOVA and generalized linear mixed models. Indolent NHL scores were compared to norm general population scores. A total of 52 patients were identified, of which 46 (88%) remained on active surveillance at 36 months. There was no significant change in any of the FACT-LYM scores over 36 months. As compared to the general population, indolent NHL patients had higher, clinically-meaningful scores in physical, functional, and social well-being, but not emotional well-being. Patients with indolent NHL on active surveillance have globally preserved HRQoL for up to 3 years after diagnosis. Emotional well-being continues to be an unmet need during active surveillance.

PRECIS

Patients with treatment-naive indolent NHL on active surveillance have preserved overall quality of life for up to 3 years after diagnosis, with higher, clinically-meaningful scores in physical, functional, and social well-being domains as compared to the general population. However, emotional well-being continues to be an unmet need for patients on active surveillance.

INTRODUCTION

Indolent non-Hodgkin lymphomas (NHL) are incurable chronic lymphoid neoplasms that are generally treated when a patient is symptomatic or develops end-organ complications from their lymphoma. Previous studies have not demonstrated a difference in overall survival between an active surveillance strategy, also known as “watch and wait,” and an immediate treatment strategy in asymptomatic patients without end-organ dysfunction.^1-3 Patients with indolent NHL managed with active surveillance may have a prolonged period between diagnosis and therapy initiation, with 46% of patients in the watch-and-wait cohort of a follicular lymphoma randomized trial not needing any treatment at 3 years after diagnosis.² As such, active surveillance is considered an appropriate strategy for asymptomatic patients with indolent NHL and is endorsed in several national lymphoma guidelines.^4-6

However, active surveillance, often re-labeled “watch and worry” by patients, may adversely affect patient-reported health-related quality of life (HRQoL) as symptom burden increases.^7-9 Data from several observational studies have been mixed,^10,11 with some studies demonstrating poor HRQoL compared to the general population, particularly in domains of physical, cognitive, and emotional functioning,^12,13 and others finding no difference or even better HRQoL scores than the general population.^13,14 Data are also mixed with respect to differences in HRQoL between patients on active surveillance versus those on treatment.^10,15 In the largest randomized controlled trial of active surveillance in indolent NHL that incorporated HRQoL as a primary endpoint, no difference was found in HRQoL between patients with follicular lymphoma managed with active surveillance versus those who received induction rituximab over the first 7 months of the study.² However, many of these observational studies were cross-sectional and did not assess individual patient-level changes in HRQoL longitudinally over time, which is particularly relevant given that patients with indolent NHL may be under active surveillance for many years prior to initiation of therapy. Further, these studies were not conducted in the era of modern therapies, and patients under active surveillance may report unique symptom burdens impacting HRQoL when faced with initiating future treatment with novel oncolytics that may have improved toxicity and side effect profiles compared to traditional chemotherapy.

The impact of an active surveillance strategy on long-term HRQoL in untreated NHL patients is not well understood and data describing the effect of active surveillance on HRQoL will better inform patients on their expectations for HRQoL when watch-and-wait is recommended. As such, the purpose of this study was to longitudinally examine patient-reported HRQoL in patients with indolent NHL managed with active surveillance in the first 3 years after diagnosis, utilizing a prospectively-collected database of patient-reported outcomes (PROs). We hypothesized that overall HRQoL would be longitudinally stable for patients who remained on active surveillance. Additionally, HRQoL scores of lymphoma patients on active surveillance were compared to established general population norm scores of HRQoL.

METHODS

Study Population

The Hoogland Lymphoma Biobank prospectively enrolls all adult patients who present for evaluation of lymphoma to three cancer centers within the University of Chicago Medicine network. Patients in the biobank who are enrolled within 6 months of their initial lymphoma diagnosis are sent a HRQoL questionnaire at 6, 12, 18, 24, and 36 months after enrollment. Patients are only sent follow-up HRQoL questionnaires if they completed the previous HRQoL assessment; as such, all HRQoL data in the biobank represents serial individual patient-level surveys. All patients with indolent NHL (follicular lymphoma [FL], marginal zone lymphoma [MZL], chronic lymphocytic leukemia [CLL], small lymphocytic lymphoma [SLL], and lymphoplasmacytic lymphoma [LPL]) in the Hoogland Lymphoma Biobank who completed at least the 6 month HRQoL survey prior to 2020 were screened for inclusion in this study. Patients were included in the study if they were not receiving any therapy at their first HRQoL survey at 6 months after enrollment, and all subsequent follow-up surveys were included at 12, 18, 24, and 36 months if patients were still untreated at those timepoints. Patient demographics, comorbidities, and current therapy are routinely collected at each follow-up point within the biobank. This study and the biobank were approved by the University of Chicago Institutional Review Board.

HRQoL Instruments and Measurements

HRQoL is assessed by the Hoogland Lymphoma Biobank with the multi-dimensional 42-item Functional Assessment of Cancer Therapy – Lymphoma (FACT-LYM) instrument,¹⁶ which assesses total general HRQoL by utilizing sub-domains of physical well-being (PWB), social well-being (SWB), emotional well-being (EWB), functional well-being (FWB), and lymphoma-specific symptoms (LYM). Higher scores on the overall instrument as well as each sub-domain indicate a better HRQoL. Several overall HRQoL summary scores can be calculated from the FACT-LYM instrument, including the FACT-General (FACT-G) total summary score (FACT-G TOT), FACT-LYM Trial Outcome Index (FACT-LYM TOI), and FACT-LYM total summary score (FACT-LYM TOT).¹⁶ In addition, single-item reliable and valid assessments of fatigue, GP1 (“I have a lack of energy”) and BMT6 (“I get tired easily”), are available for analysis. The FACT-LYM instrument has been validated in patients with NHL,^16,17 and a minimally important difference (MID) has been estimated as a change of at least 2-3 for the subscores (PWB, SWB, EWB, FWB) and at least 3-7 for the FACT-G TOT.¹⁸ When comparing group mean scores, the low end of the MID range is used to interpret group differences.¹⁹ The FACT-LYM scoring schema accounts for missing data by prorating using the average of the other answers in a particular scale.

Statistical Analyses

For the primary aim to determine if the HRQoL of patients with indolent NHL managed with active surveillance changes longitudinally from 6 to 36 months after diagnosis, all of the FACT-LYM summary scores and subscores at 6, 12, 18, 24 and 36 months were analyzed by one-way repeated measures ANOVA (rANOVA) with Greenhouse-Geisser correction for sphericity, as well as generalized linear mixed modeling (GLMM) with random intercepts and random slopes. Single-item assessments of fatigue were also analyzed over 36 months by both rANOVA and GLMM. For the GLMM, age, gender, race, and presence of comorbidities were included as non-time-varying coefficients in the models.

The secondary aim sought to compare HRQoL scores of NHL patients on active surveillance with norm scores for the general adult population and the ambulatory adult cancer population, which have been published in the literature.²⁰ In addition, HRQoL scores were compared with the subset of the ambulatory adult cancer population with an Eastern Cooperative Oncology Group (ECOG) performance status of 0, as it was hypothesized that patients with indolent NHL on active surveillance would also have an ECOG of 0. Mean subscores and total summary scores (with the exception of LYM, FACT-LYM TOI, and FACT-LYM TOT which do not have established population norms) at 6, 12, 18, 24 and 36 months were compared with the aforementioned norm scores utilizing two-sided t-tests with unequal variances. All analyses were performed in statistical software STATA Version 15.1 (StataCorp LLC, College Station, TX).

RESULTS

A total of 52 patients with untreated indolent NHL were identified in the Hoogland Lymphoma Biobank, with patient and disease characteristics displayed in Table 1. The median age of the cohort was 62 (range, 36-81) with 54% identifying as women. The majority of patients were white (88%) with 6% identifying as Hispanic or Latino. The majority of the cohort (70%) had at least 1 reported comorbidity. The majority of patients had been diagnosed with FL (54%), followed by CLL/SLL (25%), MZL (19%), and LPL (2%). The majority of FL, MZL and LPL patients had advanced-stage disease (56%), and the majority of CLL/SLL patients had Rai Stage I disease (69%). A total of 64% (33/52) of patients completed surveys at 12 months, 52% (27/52) at 18 months, 37% (19/52) at 24 months, and 19% (10/52) at 36 months after enrollment. Forty-three of the 52 patients (83%) had complete clinical follow-up data at 36 months, whereas 9 (17%) did not have complete clinical follow-up data at 36 months. Six patients in the cohort (12%) eventually required treatment for their lymphoma, at a median of 13 months after enrollment (range 7.4-64.1 months).

Table 1.

Patient and indolent NHL characteristics.

Patient Characteristics	N = 52
Age
Median	62
Range	36-81
Gender
Male	24 (46%)
Female	28 (54%)
Race
White	46 (88%)
Black	4 (8%)
Asian	1 (2%)
Multiracial	1 (2%)
Ethnicity
Hispanic or Latino	3 (6%)
Presence of at least 1 comorbidity	36 (70%)
ECOG
0	46 (88%)
1	6 (12%)
Type of Lymphoma
FL	28 (54%)
CLL	13 (25%)
ENMZL	3 (6%)
SMZL	4 (8%)
NMZL	3 (6%)
LPL	1 (2%)
Ann Arbor Stage (n = 39)
I-II	17 (44%)
III-IV	22 (56%)
Rai Stage (n = 13)
0	3 (23%)
I	9 (69%)
II	1 (8%)
Presence of B symptoms	4 (8%)
Presence of extranodal disease	7 (14%)
Presence of bulky disease	1 (2%)
Eventually required lymphoma therapy	6 (12%)
Time until therapy initiation (n = 6, 12%)
Median (months)	13
Range	7-64
Type of therapy (n = 6, 12%)
weekly rituximab x 4	2
BR x 6	2
ibrutinib	2
Number of QOL surveys at each timepoint
6 months	52 (100%)
12 months	33 (64%)
18 months	27 (52%)
24 months	19 (37%)
36 months	10 (19%)

Longitudinal Changes in HRQoL Over 36 Months

Means and standard deviations for the FACT-LYM summary scores and subscores at all timepoints, as well as the rANOVA analysis, are displayed in Table 2. By rANOVA, there was no significant change in any of the summary scores or subscores over 36 months.

Table 2.

Mean and standard deviations of PWB, SWB, EWB, FWB, LYM, FACT-G TOT, FACT-LYM TOI and FACT-LYM TOT scores over 36 months, as well as p-values for one-way repeated measures ANOVA (rANOVA) over 36 months.

	6 months (n=52)		12 months (n=33)		18 months (n=27)		24 months (n=19)		36 months (n=10)		rANOVA p-value
Subscore	mean	stdev	mean	stdev	mean	stdev	mean	stdev	mean	stdev
Physical (PWB)	25.1	4.4	25.5	2.2	25.4	2.6	26.2	2.2	25.5	2.5	0.76
Social (SWB)	23.1	4.5	22.8	4.0	23.4	4.4	23.7	4.8	23.5	4.0	0.42
Emotional (EWB)	19.7	3.3	20.0	3.0	19.3	5.0	20.2	4.2	20.1	2.0	0.67
Functional (FWB)	23.0	4.2	23.2	4.8	22.4	4.3	23.4	4.7	22.5	6.3	0.19
FACT-G TOT	90.9	11.4	91.5	10.9	90.5	11.5	93.5	11.6	91.6	12.7	0.40
Lymphoma-specific (LYM)	50.7	6.9	51.7	7.2	51.6	6.2	52.0	8.7	47.0	18.5	0.26
FACT-LYM TOI	98.9	13.2	100.5	12.9	99.4	11.5	101.6	12.3	95.0	21.2	0.19
FACT-LYM TOT	141.6	17.0	143.2	17.0	142.0	16.6	145.5	17.7	138.6	22.9	0.32

When analyzed using a GLMM model as shown in Table 3, there were no significant changes in HRQoL for PWB, FWB, FACT-G TOT, LYM, FACT-LYM TOI, or FACT-LYM TOT over time. However, the random slopes models for SWB and EWB were statistically and clinically significant. For SWB, patients with no comorbidities had a lower SWB at 6 months after diagnosis which improved over 36 months, whereas patients with baseline comorbidities had a higher SWB at 6 months after diagnosis which was stable over 36 months. For EWB, women had a lower EWB at 6 months after diagnosis which improved over 36 months, whereas men had a higher EWB at 6 months after diagnosis which was stable over 36 months.

Table 3.

Generalized linear mixed models with random intercepts and random slopes to estimate changes in FACT-G and FACT-LYM subscores and summary scores over 36 months, including age, sex, race, and comorbidities as non-time-varying coefficients in the models.

		age		gender		race		comorbidities		random effects			P-value for model
Subscore	Model	Estimate (SE)	p-value	Estimate (SE)	p-value	Estimate (SE)	p-value	Estimate (SE)	p-value	constant	time	residual
Physical (PWB)	Intercept	0.02 (0.06)	0.69	−1.65 (1.14)	0.15	−1.30 (1.66)	0.43	−1.04 (1.16)	0.37	12.54 (2.95)		2.75 (0.43)	0.60
	Slope (time)	0.023 (0.06)	0.69	−1.65 (1.14)	0.15	−1.30 (1.66)	0.43	−1.04 (1.16)	0.37	12.54 (2.95)	0	2.75 (0.43)	0.60
Social (SWB)	Intercept	0.06 (0.06)	0.34	−1.56 (1.20)	0.20	2.78 (1.77)	0.12	2.90 (1.23)	0.02	13.33 (3.10)		4.59 (0.69)	0.10
	Slope (time)	0.07 (0.06)	0.23	−1.59 (1.20)	0.19	2.95 (1.75)	0.09	3.24 (1.22)	0.008	12.48 (3.09)	0.28 (0.29)	4.05 (0.67)	0.040
Emotional (EWB)	Intercept	0.05 (0.05)	0.24	1.45 (0.86)	0.09	1.21 (1.32)	0.36	0.51 (0.90)	0.57	4.91 (1.51)		7.19 (1.03)	0.18
	Slope (time)	0.06 (0.04)	0.18	1.72 (0.82)	0.04	0.88 (1.23)	0.48	0.73 (0.85)	0.39	2.64 (1.70)	0.48 (0.33)	6.64 (1.03)	0.048
Functional (FWB)	Intercept	0.04 (0.07)	0.52	−1.05 (1.27)	0.41	0.28 (1.87)	0.88	1.44 (1.30)	0.27	15.51 (3.44)		3.98 (0.59)	0.54
	Slope (time)	0.04 (0.07)	0.51	−1.04 (1.26)	0.41	0.32 (1.85)	0.86	1.50 (1.29)	0.25	14.99 (3.64)	0.06 (0.17)	3.92 (0.62)	0.54
FACT-G-TOT	Intercept	0.19 (0.17)	0.27	−2.65 (3.27)	0.42	3.49 (4.79)	0.47	4.26 (3.33)	0.20	103.42 (22.75)		23.26 (3.48)	0.57
	Slope (time)	0.20 (0.17)	0.23	−2.35 (3.22)	0.47	3.35 (4.68)	0.47	4.75 (3.26)	0.15	95.62 (22.26)	1.58 (1.43)	20.78 (3.58)	0.45
Lymphoma-specific (LYM)	Intercept	0.12 (0.11)	0.27	1.07 (2.05)	0.60	−3.81 (3.11)	0.22	1.34 (2.13)	0.53	30.46 (8.79)		33.75 (4.91)	0.24
	Slope (time)	0.12 (0.10)	0.23	0.99 (1.96)	0.61	−3.25 (2.91)	0.26	0.69 (2.01)	0.73	19.90 (8.72)	2.39 (1.26)	29.16 (4.60)	0.34
FACT-LYM TOI	Intercept	0.20 (0.20)	0.32	−1.69 (3.83)	0.66	−4.53 (5.64)	0.42	1.95 (3.91)	0.62	136.10 (31.77)		44.96 (6.77)	0.38
	Slope (time)	0.19 (0.20)	0.34	−1.48 (3.83)	0.70	−3.86 (5.58)	0.49	1.24 (3.89)	0.75	128.55 (31.66)	3.47 (2.22)	37.59 (6.41)	0.56
FACT-LYM TOT	Intercept	0.32 (0.25)	0.21	−1.72 (4.92)	0.73	−0.04 (7.22)	0.99	5.76 (5.01)	0.25	229.54 (51.30)		62.36 (9.31)	0.35
	Slope (time)	0.33 (0.25)	0.19	−1.40 (4.86)	0.77	0.24 (7.08)	0.97	5.56 (4.94)	0.26	213.36 (50.51)	4.13 (3.07)	54.78 (9.13)	0.38

For the analysis of fatigue, there was no significant change in the GP1 (p=0.52) or BMT6 (p=0.10) scores over 36 months by rANOVA. Using the aforementioned GLMM model, there were also no significant changes in fatigue over time.

Comparison with Population Norms

Reference population norm scores for the United States general adult population, ambulatory cancer population, and ambulatory cancer population with ECOG of 0 are summarized in Supplementary Table 1.

As shown in Figure 1, lymphoma patients on active surveillance had statistically higher and clinically-meaningful PWB, FWB and FACT-G TOT scores as compared to the general population and general cancer population at each timepoint from 6 through 36 months, with the exception of FWB at 36 months due to a low number of surveys at that timepoint. However, there were no clinically meaningful differences in PWB, FWB or FACT-G TOT when compared to the ECOG 0 cancer population.

Figure 1.

Mean scores of Physical Well-Being (PWB), Functional Well-Being (FWB), Emotional Well-Being (EWB), Social Well-Being (SWB), and FACT-G total summary score (FACT-G TOT) of the active surveillance lymphoma cohort (black), general population norm cohort (blue), cancer population norm cohort (red), and ECOG 0 cancer population norm cohort (orange) at each timepoint from 6 through 36 months. Higher scores indicate a better quality of life. Statistically-significant differences between the lymphoma cohort and each population norm cohort is indicated with asterisks. Clinically-significant differences between the lymphoma cohort and each population norm cohort, based on the minimally important difference for each subscale, is indicated with a plus sign.

On the contrary, the EWB of lymphoma patients was not found to be statistically different from the general population or the ECOG 0 cancer population at any timepoint. Although it was statistically higher than the general cancer population at 6, 12, 24, and 36 months, this difference was not consistently clinically meaningful.

The SWB of lymphoma patients was statistically and clinically higher than the general population at 6 through 36 months, although it was not significantly different from the general cancer population or the ECOG 0 cancer population. It is important to note that the SWB of the general population norm score (Supplementary Table 1) is statistically lower than that of the adult cancer population reference (p>0.0001).

DISCUSSION

In this study of prospectively-collected patient-reported outcomes data, HRQoL of patients with indolent NHL managed with active surveillance did not change over the 3 years after diagnosis, with preserved HRQoL in all domains measured by the FACT-LYM instrument, including fatigue. Further, HRQoL scores for lymphoma patients on watchful waiting were significantly higher than the general population in physical, functional, and social domains, although not in the emotional domain. Our results are consistent with previous studies comparing untreated NHL patients with population norms, with this longitudinal study adding that patients on active surveillance may enjoy long-term, preserved overall HRQoL over time without significant deterioration across various HRQoL domains.

The three-year longitudinal nature of this study adds important data to the current body of evidence, as few studies have explored longitudinal changes in HRQoL in untreated indolent NHL managed with active surveillance. In a large cross-sectional cohort of 1482 patients with CLL/SLL (60% who were untreated), there was no change in emotional well-being based on time since diagnosis, although the analysis was not performed with serial patient-level surveys as in our study.¹³ In one of the few longitudinal studies, 76 patients with CLL/SLL (56% untreated) were administered serial surveys over 12 months and found to have stable HRQoL on the EORTC QLQ-C30 instrument.²¹ In another longitudinal study of 1050 patients with indolent lymphomas (45% untreated), overall HRQoL on the FACT-G was also found to be stable at 3 years of follow-up compared to baseline, although emotional well-being significantly improved at a clinically-meaningful threshold over that time period.42 In that study, 20% of patients had a baseline PRO survey collected after initiation of therapy, whereas no patients in our study had received any treatment at the time of initial PRO survey collection or at any subsequent timepoints; the inclusion of patients who had received treatment in that study may account for different findings with respect to emotional well-being, or may be due to the smaller sample size of our study. There have been mixed findings in other types of indolent NHL other than CLL/SLL, with stable physical and functional well-being but improved emotional well-being over 7 months in the active surveillance arm of a randomized trial in FL,² but worsening global, social and functional well-being and stable emotional well-being over 1 year in a prospective survey study of patients with all types of indolent NHL.²² These discrepancies likely reflect different patient populations studied as well as the importance of controlling for patient and disease characteristics when analyzing PROs. As an example from our study, rANOVA did not demonstrate any significant changes in HRQoL over time, but GLMM modeling with age, gender, race, and presence of comorbidities as covariates did reveal significant and clinically-meaningful differences in social and emotional well-being over time, with the former driven by presence of comorbidities and the latter by gender. Previous research has demonstrated that social support is protective of HRQoL in not only cancer patients^23,24 but also in patients with non-cancer comorbidities;^25,26 as such, our finding that patients with baseline comorbidities initially had higher social well-being than those without comorbidities suggests better social capital for patients in our cohort with pre-existing comorbidities. This may be due to the increased social support that cancer patients receive after their diagnosis or compensation for low emotional well-being with higher social support.^20,27,28 Poorer emotional well-being in women as compared to men with cancer has been previously demonstrated in the literature, including in lymphoma survivors,^29-31 and highlights the external validity of our cohort. Overall, this study suggests that there may be specific subgroups of patients with indolent NHL on active surveillance who are at risk for inferior HRQoL in specific domains, with dedicated prospective study of these subgroups in larger cohorts warranted.

Our finding that patients with indolent NHL on active surveillance have higher, clinically-meaningful scores in several HRQoL domains as compared to the general population suggests that patients on watchful waiting may be reassured that their HRQoL may be generally better than expected in the short- and long-term. Improved HRQoL in patients with untreated indolent NHL as compared to the general population has been previously demonstrated in some studies,^2,13 although others have demonstrated no significant difference^14,32,33 while others have found worse global HRQoL.^12,34 The longitudinal observation that physical, functional and social well-being scores were higher at both statistical and clinically-meaningful thresholds additionally highlights the importance of minimally-important differences (MID) in interpreting PROs. Utilizing cancer patient norm population scores as a reference population in this study illustrates that not all statistical differences are clinically meaningful to patients. For example, although the indolent NHL cohort had statistically higher scores in physical and functional well-being as compared to the ECOG 0 cancer population at several timepoints (Figure 1), the numerically higher scores when compared to the general population and cancer population norm scores (which exceed the MID) clearly illustrate clinically meaningful differences. MID cutoffs for various PRO instruments additionally enables comparison between studies. As an example, although one longitudinal study of patients with CLL/SLL found statistically lower HRQoL in untreated patients compared to the general population, these differences were not clinically meaningful based on the MID.¹²

In contrast, emotional well-being is consistently found to be worse among patients with indolent NHL on active surveillance as compared to the general population,^2,13,34 with our study finding no difference. This finding was surprising given that all the other FACT-G subscores were significantly higher than the general population, and suggests emotional well-being as a potential unmet need for patients on active surveillance. Inferior emotional well-being in CLL/SLL patients was initially demonstrated by Shanafelt et al. in 2007, in which emotional well-being was found to be significantly worse than both the general population and the general cancer population, with lower EWB scores of 16.8-17.8 that were clinically meaningful compared to reference norm scores.¹³ Those scores were worse than the emotional well-being scores of 19.3-20.2 in our cohort, which likely reflects differences in patient populations, study design, and the overall treatment landscape. Our additional longitudinal finding that emotional well-being is stable over 3 years, as documented in other cross-sectional studies based on time since diagnosis,^13,35 suggests that patients with indolent NHL do not emotionally acclimate to their diagnosis and to living with their disease while under active surveillance. This may be due to a feeling that “nothing is being done”¹³ or poor caregiver and social support causing maladaptive coping,³⁶ and suggests that patients on active surveillance should be closely monitored for poor emotional well-being and distress with a low threshold for additional mental health counseling, as is recommended by National Comprehensive Cancer Network (NCCN) guidelines.43 Systematic collection of PRO surveys in patients on active surveillance in routine clinical practice may improve patient satisfaction, clinician awareness of new or worsening symptoms, and global HRQoL in patients with a chronic and incurable lymphoma.³⁷ Future studies should administer prospective PRO instruments to a much larger and more diverse cohort of patients on active surveillance, potentially with the addition of a psychological support intervention; previous studies of lifestyle and counseling interventions in patients with prostate cancer on active surveillance have shown improvements in mental health-related quality of life.^38,39

There are inherent limitations to this study as an analysis of a database of PROs, including the observational nature of the study, enrollment of patients who presented only to a single center, selection bias with poor racial and ethnic representation, and a relatively small cohort size that is suboptimally powered for the GLMM analysis, although the sample size is comparable to other indolent NHL HRQoL studies. A significant limitation of this study is attrition in survey responses after 18 months of follow-up, likely due to survey administration by mail as well as the COVID-19 pandemic; future prospective studies should explore electronic survey administration of PROs. In addition, the Hoogland Lymphoma Biobank does not collect a HRQoL survey at the time of baseline enrollment, although the first survey collected at 6 months after enrollment still represented untreated patients on active surveillance. Our study did not include other important socioeconomic predictors of HRQoL, such as educational achievement, marital status, or type of insurance, as financial toxicity is a known predictor of HRQoL in patients with cancer and specifically NHL.^40,41 Strengths of this study include the analysis of longitudinal, real-world, prospectively-collected data with the collection of serial patient-level validated HRQoL surveys, which enabled mixed models analysis. Further, this study has a long follow-up compared to other prospective HRQoL studies and includes underrepresented indolent NHL subtypes in the literature such as MZL.

In conclusion, our longitudinal study suggests that patients with indolent NHL on active surveillance have preserved HRQoL for up to 3 years after diagnosis, with clinically-meaningful increases in physical, functional, and social well-being scores as compared to the general population. Given the lack of overall survival benefit with active surveillance versus immediate treatment in indolent NHL, our data continues to reinforce that active surveillance is an appropriate upfront strategy that preserves overall HRQoL. However, emotional well-being continues to be an unmet need for patients on active surveillance, and patients may benefit from early distress or mental health screening after diagnosis. Further prospective study of subgroups who may require additional supportive care while on active surveillance is warranted, particularly in larger patient cohorts.