Quality of life and survivorship in patients with low-grade ovarian cancer.

Participants were at least 18 yo with histologically confirmed primary epithelial carcinoma of the ovary, fallopian tubes, or peritoneum. Women were recruited between 2003 and 2020 from two tertiary midwestern and one southeastern university hospitals as part of a larger OC biobehavioral study. Participants were excluded if they had tumors of low malignant potential, use of systemic corticosteroid medications within the last month, pregnancy, medical conditions with known immune system impacts, another primary site cancer within the last 5 years, or cognitive impairment preventing answering surveys. Women receiving either upfront surgery or neoadjuvant chemotherapy (NAC) were eligible. All procedures were approved by Institutional Review Boards of participating institutions. The final sample included 578 HGOC patients and 85 LGOC patients. Demographic information was self-reported. Clinical information from medical records included age, stage, grade (high versus low), body mass index (BMI), histology, CA 125 level, debulking status and neo- or adjuvant chemotherapy. Grade was determined from pathology reports; slides interpreted as grade 2 before the change to the two-tier system were reclassified by gynecologic pathologists[ 3 ]; two equivocal cases were excluded. Women with masses suspicious for OC were recruited and consented during a pre-surgical visit. Psychosocial surveys were completed between the pre-surgical visit and surgery or NAC for baseline measurements, and at 6-month and 1-year follow-up visits. Peripheral blood was collected in the morning prior to surgery or to NAC, and at the 6-month and 1-year follow-up visits. Blood samples were centrifuged and then frozen at −80× C. For three days prior to surgery or NAC, and at the 6-month and 1-year follow-up visits, salivary cortisol samples were collected upon awakening (4–9 am), in the late afternoon (4–6 pm) and before bed (8–12 pm). To reduce subject burden, a subset of patients (n=98) only collected cortisol twice, upon awakening and at bedtime. Participants were requested to refrigerate self-labelled cortisol samples until brought to the clinic- for storage at −80°C until analysis[ 20 ]. The Center for Epidemiological Studies-Depression Scale (CES-D) is a well-validated 20 item self-report questionnaire in which participants record how often they have felt symptoms of depression in the past 7 days[ 21 ]. The CES-D vegetative subscale examines vegetative, primarily somatic symptoms as a component of depression. These include fatigue, anorexia, anhedonia, low motivation, and decreased desire for social interaction. The depressive affect scale examines primarily mood related components of depression. A total score of ≥16 is consistent with moderate and a score of ≥24 is consistent with severe depressive symptomatology[ 21 ]. The Social Provisions Scale (SPS) is a 24-item self-report scale that measures perceived social support[ 22 ]. The six subscales of the SPS include: reliable alliance, reassurance of worth, social integration, guidance, attachment and nurturance. The general population has been reported to have SPS scores in the range of 78–82[ 22 ]. The Pittsburgh Sleep Quality Inventory (PSQI) is a 19-item self-report scale assessing perceived sleep quality over the past month[ 23 ]. A Global Sleep Disturbance score of 5 is consistent with sleep dysfunction, and a score of 10 or higher is consistent with clinically impaired sleep. The Functional Assessment of Cancer Therapy (FACT-G) is a 27-item questionnaire that assesses four domains of QOL well-being in cancer patients: physical, social, emotional, and functional [ 24 ] and additionally includes a 12-item ovarian-specific well-being score (FACT-O)[ 25 ]. Differences of 5 points in the total FACT score indicate clinically meaningful differences; differences of 2 points on the subscales represent meaningful differences. The mean FACT-G score for the U.S. population is reported to be 80.1[ 26 ]. The Profile of Mood States- short form (POMS-SF) is a 37-item self-report questionnaire assessing dimensions of affect[ 27 ]. POMS assesses the following mood domains on a 5- point Likert scale from (0) not at all to (4) extremely: anxiety, depression, anger, vigor, fatigue, and confusion. The relevant subscale for this study was the fatigue subscale. The POMS-SF has been validated against the CESD and the POMS-long form to assess mood and distress in cancer patients[ 28 ]. The Perceived Stress Scale is a 14-item well-validated self-report questionnaire measuring individual subjective stress levels[ 29 ]. The survey asks how often an individual thought or felt a certain way from (0) never to (4) very often. Maximum score is 56. A national sample of community living females (n=1406) had mean scores of 20.2 (S.D.=7.8).[ 30 ] Plasma interleukin-6 (IL-6) was analyzed by ELISA (R&D Systems, Minneapolis, MN) with results interpolated from the standard curve. The minimum detectable level of the assay is < 0.7 pg/mL and inter-assay coefficient of variance ranged from 3.3–6.4%. Samples with an intra-assay coefficient of variance over 20% were excluded. IL-6 samples below the sensitivity of the regular assay were assessed with the R&D Systems High Sensitivity (HS) ELISA. Twenty-seven plasma samples with values under the detectable limit could not be assayed using the HS kit and were assigned the lowest detectable value (0.7 pg/mL). To normalize the distribution of values, a log (10) transformation was performed. Salivary cortisol was assayed using a commercial chemiluminescence immunoassay (IBI, Hamburg, Germany). The lower detection limit is 0.41 nmol/L and coefficients of variance are < 10% for both inter-assay and intra-assay assessments. To normalize the distribution of values, a natural log transformation was performed. A typical diurnal cortisol pattern in a healthy subject displays a cortisol peak in the morning, with a nadir at night[ 31 ]. Because our prior work showed nocturnal cortisol and cortisol slope are related to survival [ 19 ], we focused on these two facets of diurnal cortisol. LCMS-based quantitation of cortisol from ovarian tissue samples was performed at Columbia University by liquid-liquid extraction followed by gradient separation on Waters Aquity UPLC BEH C18 column (1.7μm, 2.1×50mm) and analyzed on a Waters Xevo TQS MS with an ACQUITY UPLC system (Waters, Milford, MA, USA) under positive ESI with MRM mode. The assay has an intra-assay variation of 3% and inter-assay variation of 2.6%. To remove outliers in the cortisol data, cortisol sampling time ranges were based on personal awakening times, which is associated with a rise in cortisol[ 32 ]. The ranges were designed to fit the maximum number of participants while maintaining homogeneity. Samples were excluded if they were more than four standard deviations above the mean of their specific time points. Mean cortisol values were determined for the awakening and nocturnal time points by averaging cortisol levels obtained over three days for each participant, a previously validated method [ 20 , 31 ]. Individual cortisol slopes were determined through regression of cortisol values obtained from the data pooled over the 3 collected days[ 32 ]. Slopes derived from 3 time point collections (n=164) did not differ significantly from slopes derived from 2-point collections (n=98) ( p =0.523). Because of potential confounds, patients who were receiving chemotherapy or parp inhibitors at the time of the one-year assessment were not included in biomarker analyses. Analyses were performed using the Statistical Package for the Social Sciences v. 29 (SPSS, IBM, Armonk, NY) and RStudio version 2023.03.0 using R version 4.2.2. Descriptive statistics were used to summarize demographic and clinical characteristics. Independent samples T-tests were used to compare means, chi-square was used to compare greater than two nominal variables and Fisher’s Exact Test was used to compare two nominal variables. Distributions were examined for normality and outliers, and transformations applied as appropriate. A priori covariates (age and cancer stage) were included in all analyses. Multi-level modeling incorporating all three time points (baseline, 6 months, 1-year) were used to examine trajectories. Each model was fit using a participant intercept and slope term; reduced models using fixed slope and participant intercepts were tested and used when appropriate by model fit. General linear models were used to examine differences in tumor cortisol between low- and high-grade patients. A p -value of 0.05 was statistically significant.

Participants’ mean age was 62 with a standard deviation of 11 years. LGOC patients were younger than those with HGOC ( p <0.001) and had higher BMI ( p =0.028). Participants were predominantly non-Hispanic white. Most were living with a partner or spouse in urban areas and had completed post-secondary education. LGOC patients presented predominantly with stage I or II disease, while the majority with HGOC had advanced disease. Most HGOC patients had histology suggesting serous subtype (80.8%); whereas only 31.8% of LGOC were of serous sub-type histology. Among LGOC patients, slightly over one third had endometrioid histology (39.5%) and without consistent mention of concurrent endometriosis pathologically. HGOC patients had a significantly higher proportion of ascites ( p =0.008). Mean plasma CA125 level in LGOC patients was significantly lower than in HGOC ( p= 0.028). Consistent with practice at the time of the study, approximately 14% of patients in either group received NAC ( Table 1 ) and per NCCN guidelines, patients with stage III/IV LGOC and HGOC received platinum and taxane-based chemotherapy. As seen in Table 2 , at the time of diagnosis, the mean total CES-D score in LGOC (10.85±3.99, p =0.044) was significantly lower than HGOC patients (13.35±2.35) although both means were still in the non-depressed range. Compared to LGOC (40.7%), a greater proportion of HGOC patients (47.7%) reported depressive symptomatology in the moderate or severe clinical range (CESD ≥16) at study entry; however, this difference was not significant ( p= 0.39). Patients with LGOC reported significantly less symptoms of vegetative depression ( p =0.018), but not of depressive affect than those with HGOC ( p =0.111). Patients with LGOC and HGOC reported moderate sleep disturbances with average scores above the clinical range for disturbed sleep. Sleep quality of LGOC patients (6.35±1.93) was significantly better than that of HGOC patients (7.67±1.12) ( p =0.014). More HGOC patients (75%) had sleep dysfunction (PSQI ≥5) and clinically disturbed sleep (PSQI ≥10) (33.8%) than those with LGOC (70%; 21.7% respectively); however, the difference in proportions of HGOC and LGOC patients with sleep disturbances was not statistically significant ( p =0.167). There were no differences in fatigue according to grade ( p= 0.102). LGOC or HGOC patients reported strong social support with no difference in perceived social support by grade ( p= 0.26). Mean scores for both grades approximated levels of perceived stress reported in a community sample of 1406 females (20.2, S.D. 7.8). At study entry, approximately 44 % of LGOC and 49% of HGOC participants reported stress above these community norms. There was a non-significant trend towards LGOC patients having lower levels of reported stress than those with HGOC ( p= 0.088). ( Table 2 ). The mean QOL scores for all patients at the time of diagnosis were at or slightly below reported general population means[ 26 ], with a mean FACT-G score of 80.62±6.77 for LGOC patients and a mean of 77.59±3.66 for those with HGOC. These scores did not significantly differ by grade ( p =0.511). There was no significant difference by grade at the time of diagnosis in any facet of QOL, including the FACT-O QOL measure (all p -values ≥ 0.20). As seen in Table 2 , changes in psychosocial characteristics over the first-year post-diagnosis did not differ by grade. For example, overall depression symptoms ( p <0.001) as well as vegetative depression symptoms ( p <0.001) improved over time in both LGOC and HGOC patients, and there was no observed difference in the trajectory of change over time by grade (CES-D total: p= 0.347; vegetative: p =0.455). Mean sleep quality was still in the disturbed range for both LGOC and HGOC at one year. Sleep quality did not improve over time across both groups ( p= 0.50) and there was no difference in trajectory of change in sleep by grade ( p =0.704). Fatigue improved significantly over time ( p= 0.02) for both grades, but there was no difference in the trajectory of change by grade ( p= 0.99). There was no change in perceived stress over time ( p= 0.42) or by grade ( p= 0.55). Participants reported a decrease in total social support over time ( p= 0.03) which did not differ by grade ( p= 0.56) ( Table 2 ). LGOC and HGOC patients reported improvements in total QOL that were both clinically and statistically significant over the year following initiation of treatment ( p <0.001). The mean QOL of LGOC patients improved by almost 13 points (80.62 to 93.35) on the FACT-G scale; whereas, that of HGOC participants improved by approximately 9 points from (77.59 to 86.78). However, the differences in the trajectory of QOL improvements over time by grade were not statistically significant ( p= 0.08). At 12 months the differences in FACT means by grade were likely clinically significant (5 point difference)[ 26 ], and the QOL of the LGOC was above the population mean (80.1), whereas that of the HGOC patients was below the population mean. Examining changes in the individual QOL subscales, functional well-being (FWB) ( p <0.001) and emotional well-being (EWB) ( p <0.001) improved significantly over time for both grades; however, LGOC patients improved significantly more in their FWB ( p= 0.044) and EWB (p= 0.021) than HGOC patients. Physical well-being ( p <0.001) and FACT-O QOL ( p <0.001) both improved over time, but there was no difference according to grade for either measure. Finally, there were no changes in social well-being over time ( p= 0.77) or by grade ( p= 0.58) ( Figure 1 ). The level of nocturnal cortisol in LGOC patients was not different from that of HGOC patients, adjusting for age and stage ( p =0.414), and we did not observe a significant difference by grade in the average cortisol slope over three days ( p =0.238) ( Table 2 ). There was a possible trend towards lower mean tumor cortisol levels in LGOC compared to HGOC patients although the LGOC sample size tumors was small, and the difference was not significant ( p= 0.08) ( Table 2 ). There was no difference in plasma IL-6 between LGOC and HGOC ( p= 0.12) ( Table 2 ). Nocturnal cortisol decreased non-significantly over time during the first-year post diagnosis ( p -value = 0.12) while cortisol slope did not change appreciably over time (p=.83) and these changes did not differ by grade. There was a significant decrease in plasma IL-6 over time in both groups ( p <0.001), but the change in IL-6 over time did not differ by grade ( p= 0.12) ( Table 2 ).

We compared psychosocial characteristics of patients with newly diagnosed LGOC or HGOC. To our knowledge, this is the first characterization of psychosocial factors in patients with LGOC during the first year post diagnosis. A major finding from our research is that despite known differences in prognosis according to grade, LGOC patients did not differ from their HGOC counterparts in major survivorship issues such as depressive affect, stress, and total QOL. Although levels of depressive symptoms were significantly lower in LGOC compared to HGOC at entry, mean scores for both were in the range of mild depression. Close to half of all patients reported at least moderate depression for both LGOC and HGOC at entry, which parallels previous reports[ 14 ]. However, by one year mean scores for both groups were in the low non-depressed range.[ 14 ]. A recent study by Hu et al . [ 34 ] found that older age and distant-stage disease were associated with a greater risk for depression within five years after OC diagnosis, but our study is the first to compare levels of depression by grade. LGOC patients reported less vegetative depression symptoms than those with HGOC, but over time, vegetative symptoms decreased in both groups, potentially reflecting decreased inflammation [ 13 , 14 , 16 ] and decreased tumor burden following treatment. Most patients in our study reported sleep disturbances at baseline, which is consistent with our previous work[ 10 ]. HGOC patients had significantly poorer sleep quality than LGOC at baseline. Over time, neither group reported substantial improvements in sleep. Sustained sleep disturbances in cancer patients have been described in survivors up to nine years following treatment, and has been attributed to a variety of physical, emotional, and economic stressors in addition to a fear of recurrence[ 35 ]. Notably, lack of improvement in sleep in our participants combined with our finding that inflammation (IL-6) significantly decreased in both groups suggests that inflammation might not be a major driver of sustained sleep disturbances among LGOC and HGOC patients. Instead, poor sleep might be explained by continued pain, stress, and anxiety and the difficult experience of living with cancer[ 36 ]. Assessments of QOL, in conjunction with standard clinical measures, have been shown to better predict survival compared to use of standard clinical measures alone [ 37 ]. While overall QOL was at or below the population mean at study entry, over time, total QOL significantly improved in all patients. This improvement has been shown in several other studies of women with OC [ 10 , 38 ] and could reflect decreased tumor burden, successful treatment of cancer symptoms, acceptance, and adaptations patients have made to improve their QOL[ 39 ]. However, it is important to emphasize that QOL did not differ by grade at baseline or over time. This finding is important as it emphasizes that LGOC patients face similar levels of QOL disruption as HGOC patients. Findings from a recent multinational survey mirror our findings that LGOC patients suffer from impaired QOL and mental health disruptions. Specifically, patients with LGOC report that copays and costs, impacts on job and career, and impact on sexual health related to their cancer experience have major effects on their mental health [ 40 ]. Taken together with our findings, these survey findings indicate the importance of efforts to help LGOC patients and HGOC patients address the stressors that negatively impact their QOL. Since both peripheral and central inflammation have been shown to elicit neurocognitive symptoms such as fatigue, lethargy, and difficulty concentrating [ 13 ] we hypothesized that differences in psychosocial characteristics associated with grade might parallel grade-related differences in biomarkers of inflammation. Our previous research identified a significant relationship between symptoms of vegetative depression and elevations in both IL-6 and evening cortisol in OC patients[ 14 ]. Here, however, we did not observe differences in IL-6 or salivary cortisol according to grade at baseline. This was surprising in light of previous findings on differences in molecular characterization of inflammatory proteins in LGOC vs. HGOC [ 4 ]. This finding could be an effect of the malignant processes driving high inflammation levels in both groups, muting any meaningful difference or due to our LGOC sample being underpowered to detect the magnitude of difference. We did, however, observe a trend towards lower tumor cortisol in LGOC tumors. Previous work has shown that ovarian cancer cells express stress hormone receptors and respond to stress hormones[ 41 ]. Differences in tumor cortisol could be driven by physical and emotional stress and have the potential to impair local tumor-immune interactions[ 17 , 19 ]. Future research with larger patient numbers is necessary to better understand the complex relationship between inflammation, psychosocial experience, and patient well-being in LGOC patients. In summary, we find moderate disturbances in QOL, depression, and sleep in LGOC women, but these disturbances do not differ substantially from those in HGOC women. Similar to previous research in breast cancer patients, objective disease severity was not a robust predictor of subjective quality of life[ 42 ]. Although LGOC patients tend to be younger and have a better prognosis than HGOC patients, a reasonable proportion of these women still struggle with moderate depression and sleep disturbances at their initial clinic visit. As reported by Sun et al ., [ 40 ], LGOC patients face prolonged financial, job, sexual, health insurance, and treatment related side-effect related stressors as a result of their often-longer treatment intervals and difficulty finding the most effective treatments. We contend that QOL and mental health interventions would benefit LGOC patients. It is particularly important that physicians identify and treat depression in LGOC patients and provide longitudinal resources for support given the younger age and better prognosis [ 42 ]. Addressing sleep quality through targeted biopsychosocial interventions is likely to improve outcomes and QOL. Further, by addressing specific QOL domains, distinct and personalized survivorship paths could be created to better target the most prominent factors that negatively impact QOL. Given the low prevalence of LGOC, future research would benefit from concerted efforts to recruit larger samples of this population for study to investigate potential differences not reported here. Future research is needed to document the psychosocial experiences of LGOC patients at five- and 10-years post-treatment and to understand potential long-term impacts of current therapies on QOL.

Our findings are the first to characterize psychosocial and potentially related inflammatory factors in LGOC patients. We highlight an unmet need to address depressive symptoms, sleep quality, and QOL, both at baseline and throughout treatment, regardless of grade. Despite having better prognosis, LGOC patients have mood and sleep disturbances that impact their QOL similarly to HGOC patients. Future work is needed to map QOL and survivorship issues in LGOC patients beyond one-year post diagnosis.

The correlational study design limits our findings. The trend we observed towards higher levels of tumor cortisol in HGOC patients requires further study with a larger sample. Since LGOC patients early in the study may have been treated differently than those later in the study given the advent of letrozole as a maintenance treatment after April 2017, treatment differences could have skewed the responses and biomarker levels [ 43 ]. Some statistical comparisons may have been underpowered, as large samples of LGOC are difficult to accrue due to low prevalence and incidence. Associations and differences reported here may underestimate the impact of LGOC as compared to HGOC and should be replicated in larger samples. While we did control for age and stage as confounders in our analyses, there might be other factors that influenced our findings that we were unable to control for.

Ovarian cancer (OC) is the deadliest gynecologic malignancy. Epithelial cancers constitute 90% of all OCs, with serous ovarian carcinoma the most prevalent histology[ 1 ]. Of serous carcinomas, high-grade ovarian carcinoma (HGOC) carries the worst prognosis while low-grade ovarian carcinoma (LGOC) offers more favorable outcomes[ 1 ]. LGOC and HGOC differ in origins, chromosomal instability, copy number changes, underlying mutations, and overall progression [ 2 , 3 ] . Additionally, inflammatory pathway protein expressions differs in LGOC versus HGOC, and grade can be characterized using inflammatory molecular biomarkers [ 2 – 4 ]. Biological differences in LGOC and HGOC have led to recent innovations in precision medicine for these patients[ 5 , 6 ]. Efforts to explore factors influencing QOL in OC patients have generally focused on outcomes regardless of grade. Thus, little is known about QOL specifically in LGOC patients. Research has highlighted the important predictive value of QOL for overall survival across all OC patients[ 7 , 8 ], with sleep quality, depression, and social support as potential drivers of worse QOL. For example, OC patients have frequent sleep disturbances and poor sleep quality which are associated with depression and poor QOL[ 9 , 10 ]. Higher levels of non-cancer life stressors have also been associated with worse QOL [ 11 ]. Moreover, social isolation is associated with decreased survival in OC patients.[ 12 ]. Psychosocial attributes must be examined in LGOC patients specifically. Given the major differences in molecular biology, inflammatory proteins, treatment and prognosis between HGOC and LGOC, we hypothesized that LGOC women LGOC would have lower levels of depression, sleep disturbances, and better QOL than HGOC women. We further hypothesized that inflammatory factors may impact QOL differences. This hypothesis was influenced by the phenomenon of “Sickness Behavior” and our prior work underscoring associations between psychosocial factors and inflammation in OC[ 10 , 13 , 14 ]. Sickness behavior is the concept that inflammatory cytokines, in a state of illness, lead to behavioral changes such as anorexia, fatigue, anhedonia, and social withdrawal[ 13 ]. The pro-inflammatory cytokine IL-6 is elevated in OC patients [ 15 ] and these elevations are associated with vegetative depression, fatigue, and social isolation[ 10 , 14 , 16 , 17 ]. Cortisol, an anti-inflammatory neuroendocrine mediator and known regulator of sleep and the stress response[ 18 ], has been associated with fatigue and worse survival, specifically in the setting of nocturnal elevations and flattened cortisol slopes[ 16 , 19 ]. Thus, we explored differences in inflammatory milieu according to grade. The goal of our study was to better characterize LGOC and their concomitant psychosocial and inflammatory characteristics. To address this, we sought to 1) characterize psychosocial experiences of depression, QOL, and sleep in LGOC and HGOC patients and 2) characterize patterns of biomarkers, IL-6 and cortisol by grade.