Health-related quality of life in Hodgkin lymphoma: a systematic review

Study characteristics are presented in Table 1. Of the 65 studies, 53 (82 %) utilized cross-sectional design; only 12 (18 %) used longitudinal design. Of the 12 longitudinal studies, seven included individuals who were off treatment, while five included samples who were both on and off treatment. We identified three longitudinal RCTs [10‐12], of which patients were followed from the time of diagnosis through up to 10 years of survivorship. These studies accrued large samples and made comparison of HRQL according to randomized treatment groups. The 53 cross sectional studies reported on 35 unique cohorts and only one included patients both on and off treatment. The remaining 52 (98 %) cross-sectional studies only included individuals who were off active treatment.

The sample size of HL patients varied considerably with 78 % of studies including a sample of 50 or more HL patients. For the 53 cross-sectional studies, the median sample size was 135 (range 15–1843) and for the 12 longitudinal studies the median sample size was 51 (range 12–3208). The majority of study designs were cohort (n = 45) and 14 used a comparator cohort design. The most common comparators were the general population (n = 23), other cancer patients (n = 12) and siblings (n = 3). When comparing among other cancer patients, the most common comparator included other types of lymphoma (n = 7). The majority of studies relied on remote data collection (n = 42), when specified either by mail (n = 35) or telephone (n = 6), with the remainder conducted in person. The study funding was primarily investigator initiated (n = 58) or via a cooperative group (n = 7). We did not identify any industry-funded studies. Half (52 %) of the identified studies were published after 2005. Table 2 contains the data extraction results.

Age was reported in 54 studies (83 %). Thirty-three studies reported a mean age of 40 years (SD = 8.5) and 21 studies reported median age of 38 years (range 22–52). In 41 (63 %) studies that reported time since diagnosis, the mean was 11.8 years (n = 25) and the median was 12.4 years (n = 16). Twelve (19 %) studies reported time post treatment, with a mean of 4.8 years (SD = 1.73) (n = 4) and a median of 10.2 years (n = 3). Five studies reported only a range of years post treatment (1–23 years).

Studies varied by aspect of HRQL examined as well as by specific instrument used. Table 3 summarizes the various aspects of HRQL by study design (longitudinal vs. cross-sectional). The most commonly used multi-dimensional instruments were the EORTC QLQ-C30 (n = 18), the SF-12 or SF-36 (SF-36 Family) (n = 18) and Health Utilities Index (HUI) 2, 3 (n = 8). Strikingly, all 12 longitudinal studies included a multidimensional scale whereas only 62 % (n = 33) of the cross-sectional studies did so. Among cross-sectional studies, the most commonly used measures were the HADS (n = 8), Fatigue Questionnaire (n = 7), and the EORTC QLQ-C30 (n = 7).

Multi-dimensional HRQL scores varied by treatment type [10, 13, 14], age and sex [11], comorbidities [15] or late effects [16‐18]. Both Khimani and Adams found significantly lower quality of life among HD survivors who experienced cardiopulmonary late effects [16, 18]. Goodman and Minn found that HD patients undergoing autologous stem cell transplant reported no difference in global HRQL versus the general population, however these patients did experience a decrease in overall cognitive and social health [19, 20]. Finally, Ganz found that HRQL can improve over the trajectory of treatment into survivorship [10] and Mols supported this finding that HRQL continues to improve through long-term survivorship [21].

Among the single domains explored, psychosocial was the most frequently identified with the psychosocial assessments more common in cross-sectional (38 %) compared to longitudinal (33 %) studies. The most frequently used psychosocial instruments included the Hospital Anxiety and Depression Scale (HADS) (n = 13), the Brief Symptom Inventory Scale (BSI) (n = 6), the Profile of Mood States (POMS) (n = 5) and the Psychosocial Adjustment to Illness Scale (PAIS) (n = 5). As studies utilized a variety of scales and measures, the results within this domain were inconsistent and the presence of psychosocial distress varied among reports. Early studies indicated that HL survivors experienced increased psychological distress [22‐24], which was later supported by Oerlemans [25]; however, several others studies found no difference in the psychosocial distress of HL survivors when compared to healthy controls [26‐28].

The second most commonly assessed domain was sexual health (n = 7). When assessing sexual health, researchers used four validated sexual health instruments including the Sexual Health Scale, the Global Sexual Satisfaction Index, the Brief Sexual Function Inventory, and the Sexual Function Scale. Four of the studies indicated that HL survivors report increased SX problems [14, 22, 29, 30]. These SX problems can improve over time [11], but may be more long lasting in higher-stage patients [12].

Several of studies used symptom-specific questionnaires (n = 29). Of these symptom assessments, the 76 % majority focused on fatigue (n = 22). Other symptom assessments included nausea (n = 3), energy level (n = 1), pain (n = 1) and symptom distress (n = 1). The three most commonly used fatigue instruments included the Fatigue Questionnaire (FQ) (n = 12), the MFI (n = 5), and the Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-Fatigue) (n = 3). Results in this domain were more homogenous with multiple studies indicating that HL survivors are at increased risk for fatigue when compared to healthy controls [10, 31‐34]. Studies within this domain found correlations between fatigue and depression ([15, 32] and fatigue and cardiopulmonary late effects [18, 35]. One longitudinal study with a small sample size showed that an exercise intervention can improve fatigue in HL survivors [36].

In our review, 63 (97 %) studies used at least one validated instrument. Two exceptions reported by Abrahamsen [29] and Chao [37], used only non-validated, study-specific questionnaires. Six other studies incorporated a study specific questionnaire in addition to a validated instrument. Of these, three included broad questions including demographic, medical and psychosocial information [26, 38, 39], while the other three created assessment tools specific to their study question pertaining to cardiac health [16], patient screening [31], and psychiatric disorders [32].

Our quality assessment was based upon six criteria (Table 4). Nearly 86 % (n = 56) of all of the studies included more than two demographic variables and clinical characteristics. The majority (n = 53, 82 %) also reported quality of life scores—specifically including a mean score and standard deviation (SD) in table or text form. Of the 11 studies that did not fulfill this criteria, five presented the data graphically [10, 12, 40‐42]; three used the data to dichotomize groups (e.g., fatigued vs. non-fatigued population) [36, 43, 44]; two provided means scores but no standard deviation [41, 45]; one provided the percentages of negative impact of sexual function used to compare between groups [39], and one reported the percentage of sexual problems [29]. Most (n = 53, 82 %) of the studies compared results of individuals with HL with other groups such as siblings, general population or other individuals with cancer and 78 % of the studies included a sample size greater than 50. For the final criteria, 69 % of the studies included an analysis of HL HRQL scores stratified by HL specific characteristics.

The following section highlights specific studies with a focus on their methodological rigor. The 2013 prospective study by Behringer et al. [12] was embedded within German Hodgkin Study Group HD10-HD12 trials. This study included the largest sample within our review, 3208 patients, and followed these patients from diagnosis through up to 27 months of follow-up care. Patients were randomized to varying protocols of HL chemotherapy and radiation, according to each of the three clinical trials. This study ultimately focused on the sexual functioning (SX) domain; however it also collected multi-dimensional HRQL data. Studies instruments included the European Organization for Research and the Treatment of Cancer (EORTC-QLQ-C30), Multi-dimensional Fatigue Inventory (MFI) and a Sexual Functioning Scale. Researchers found that while SX was reduced at baseline, it improved after therapy and eventually normalized in individuals with early stage disease. Within the HD11 trial, which was a randomized comparison of doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD) versus bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, and prednisone (BEACOPP), a small but significant difference of SX symptoms was detected in favor of ABVD. Long-term SX was related more to baseline SX and patient characteristics than to the intensity of treatment. Of note, older age, advanced stage disease, and female gender had an overall negative influence on SX.

From a quality perspective, the researchers used validated, multi-dimensional assessment tools; however they did not report specific means scores or the standard deviation. Instead, they used the HRQL data as independent variables to predict/explain sexual function and reported it in graph format. This absence of summary scores and variance precludes performing meta-analysis from these published data.

In 2003 Ganz et al. [10] conducted the only RCT of HRQL in HL in the US. This study was embedded in a RCT in cooperative Southwest Oncology Group 9133 trial. It included 247 patients in a prospective, longitudinal design, evaluating multidimensional HRQL at baseline, 6 months, 1 and 2 years. Researchers found a statistical difference in HRQL between the two treatments subtotal lymphoid irradiation with combined modality treatment (CMT). Specifically, the CMT arm experienced greater symptom distress, fatigue and poorer overall HRQL; however, by years 1 to 2, patients with in the two groups did not differ in outcomes. The authors also reported that both groups experienced significantly increased fatigue at 6 months from baseline (when compared to healthy populations) and this persisted even 1–2 years after treatment.

This study used a high quality methodology with a prospective, longitudinal, RCT design, a large sample size, and validated multi-dimensional assessment tools; however, like Behringer [12], they did not report specific means scores or the standard deviation, but instead presented trends over time graphically.

Heutte et al. [11] conducted a prospective, longitudinal study assessing HRQL among patients treated on the H8 study in Europe. This study included 935 patients, who were assessed at the end of therapy and serially up 10 years following treatment using the EORTC-QLQ-C30, MFI, and sexual function scale. Their results indicated that the emotional functioning scores were more affected than the physical functioning scores, and that women reported lower HRQL and increased symptom distress than men. All of the HRQL domains they evaluated showed improvement within 18 months of treatment completion with the exception of cognitive function and motivation. The authors suggest that neither of these domains were affected by the HL or treatment. Finally, they found that high levels of fatigue at end of treatment predicted persistent fatigue into long-terma follow-up. They did not find any differences between the treatment groups. Methodologically, this study included a large xsample size, randomization between treatment groups and used validated, multi-dimensional assessment tools. Methodologic concerns included the lack of baseline data (prior to initiation of treatment) and no data obtained during treatment.

Calaminus et al. [46] conducted a cross-sectional study assessing HRQL in pediatric survivors, who were previously treated for HL in German-Austrian RCT studies from 1978 to 2002. This study enrolled 725 participants, who were assessed using the EORTC QLQ-C30. Results were compared to a similarly matched general population sample. Several of the results of this study are similar to Heutte’s: survivors experienced worse emotional and social functioning compared to normal population; females experience lower over-all functioning and higher symptom burden than men; survivors experienced greater symptoms of fatigue and sleep problems. Again, there was no relationship between the treatment types. Methodologically, this study included a large sample size, validated instruments and comparisons between groups. Although it did not follow patients over time, the study does illustrate how a well-designed cross sectional study can support longitudinal findings.

Klaassen et al. [40] examined the ability to detect change over time in four different HRQL measures. In their prospective study of 48 patients, patients were assessed at four points in time: 2 weeks after first course of chemotherapy, during the second course of chemotherapy, during their week of radiation and 1 year after diagnosis. All of the HRQL measures demonstrated significant change from Time 1 to Time 4. In a second study, Klaassen [41] also examined proxy reporting by parents and nurses of children with HL to determine if correlation with children’s report. In this study, the authors found that over the course of treatment there was statistical significant agreement among the child, parent and nurse, as measured by the Spearman rank-order correlation coefficient. Methodologically, this study used validated tools and compared trends over time; however, concerns include its small sample size, no baseline data, no randomization according to the treatment arm or disease stage and the absence of means scores with standard deviations.

Vermaete et al. [47] conducted a longitudinal study to assess fatigue, physical activity and physical fitness in individuals with lymphoma before, during and after treatment. This study included 29 patients with either HL (n = 12) or Non-Hodgkin Lymphoma (n = 17). Over course of treatment, researchers detected a decline in hemoglobin, physical force and oxygen uptake, and patients reported significantly increased fatigue. This study looks deeper into the complex relationship between fatigue, physical inactivity and deconditioning and support further work in developing exercise interventions. Methodologically, this study used validated tools, obtained baseline prior to the start of treatment and provided mean scores with standard deviation; however, it enrolled a small sample size and did not randomize according to the treatment arm or disease stage.