Modifiable pre treatment factors are associated with quality of life in women with gynaecological cancers at diagnosis and one year later : results from the HORIZONS UK national cohort study

http://sro.sussex.ac.uk Glasspool, Rosalind, Wheelwright, Sally, Bolton, Victoria, Calman, Lynn, Cummings, Amanda, Elledge, Beryl, Foster, Rebecca, Frankland, Jane, Smith, Peter, Stannard, Sebastian, Turner, Joshua, Wright, David and Foster, Claire (2022) Modifiable pre-treatment factors are associated with quality of life in women with gynaecological cancers at diagnosis and one year later: results from the HORIZONS UK national cohort study. Gynecologic Oncology. ISSN 0090-8258

Women were consented prior to receiving any treatment, except in cases where a 170 confirmed diagnosis could only be made at the time of surgery. These participants entered 171 the study following a confirmed surgical diagnosis. 172 173 Baseline questionnaires were completed by participants at study entry, with follow-up 174 questionnaires sent to participants three and 12 months later, following status checks to 175 ensure this was appropriate. Case Report Forms (CRFs) were completed by NHS site staff at 176 baseline, then six and 12 months later. A range of validated and study-specific outcome measures were selected informed by the 184 study's conceptual framework and following discussions with patient representatives and 185 clinical experts [21] (supplementary material S2). At baseline, participants provided 186 sociodemographic data (including gender, ethnicity, domestic status, household make-up, 187 level of education reached and caring responsibilities), socioeconomic data (including 188 employment, income, state benefits, pension, and accommodation) and lifestyle data 189 (including height, weight, smoking, vaping, alcohol consumption, diet and exercise). 190 Measures of participants' general health status, social support, social integration, health 191 literacy, self-efficacy, resilience and ability to navigate health services were also included.
Physiological, clinical and treatment data were collected from medical records (including 193 age, cancer type, tumour grade, cancer stage and ECOG performance status), and 194 participants' postcodes were used to obtain Indices of Multiple Deprivation. 195 196 The primary outcome is QLACS [23], which comprises 47 items grouped into 12 domains, 197 seven of which are generic and five cancer-specific. Each item is rated on a seven-point Linear regression models were constructed to examine the association between baseline 216 characteristics and QLACS-GSS at baseline (model one), QLQ-C30 summary score at baseline 217 (model two), QLACS-GSS at 12 months (model three), and QLQ-C30 summary score at 12 218 months (model four). Complete case analysis was applied in each model, i.e. only 219 participants for whom there were no missing data for the baseline characteristics and 220 outcome were included. Residual tests for normality and standard error robustness checks 221 were conducted on each model. We did not adjust for ethnicity owing to the small 222 proportion of participants who were not White British. 223 224 A p-value <0.05 was considered statistically significant. All statistical analyses were 225 performed using STATA version 15.1. 226 227 Results 228 Figure 1 shows the flow of participants through the study. All patients presenting with a 229 gynaecological cancer in the timeframe of the study were screened. Reasons for declining to 230 take part included being too busy, not wanting to take part in research, not wanting to 231 complete questionnaires, feeling too anxious or distressed or ill, privacy concerns, and not 232 wanting to be reminded of diagnosis. The demographic profile of people who declined to 233 participate was similar to those who consented. The mean age of those who declined was 234 63, and of those who consented was 61 and there was no difference in ECOG status. 235 Decliners were slightly more likely to have Stage 3 cancers (15%, as opposed to 10% of 236 consenters) and less likely to have Stage 1 cancers (69%, as opposed to 76% of consenters). 237 We cannot exclude the fact that there could have been differences in baseline and 12 238 month QOL measures between those that did and did not respond. 239

241
Insert Fig 1 about  with vulvar cancer. The majority of participants had an ECOG performance status of either 0 254 or 1, indicating good functioning. About half had at least one comorbidity which they 255 reported limited the activities they did on a typical day. Participants lived in areas from all 256 five quintiles of deprivation. Participants were scored as having lower socio-economic status 257 (SES) if they were unemployed or on benefits (excluding child benefit), were renters, or 258 lacked access to a car or to the internet. Just under half of participants reported one or 259 more of these indicators. The majority had a BMI of ≥25 indicating they were overweight or 260 obese. About half reported being physically active and most had never smoked. Most 261 participants lived with others, although less than 20% felt they were fully supported socially. 262 About a third had caring responsibilities. The majority of participants were treated with surgery (90%), 27% received chemotherapy and 21% radiotherapy. While these treatment 264 figures were not baseline characteristics, they were included in the regression models for 265 completeness. A higher proportion of women aged 50+ years at baseline went on to 266 complete the 12-month questionnaire compared to those who did not (supplementary 267 material, S3). There were no other significant differences in the baseline characteristics of 268 those who responded at 12 months and those that did not (supplementary material, S3 and 269

Model three: association between baseline characteristics and QLACS-GSS at 12 months 320
Having either one (7.25 95% CI 2.11, 12.39) or two or more (13.89 95% CI 8.33, 19.45) 321 limiting comorbidities and tumour grade three (6.80 95% CI 1.06, 12.54) were associated 322 with a higher score (worse QOL) on the QLACS-GSS at 12 months. Having a lower SES 323 (defined as an SES score of two or more (see S2)) was associated with an increase in QLACS-324 GSS of 8.85 (95% CI 2.07, 15.64) i.e. worse QOL. A one unit increase in HADS anxiety score 325 and HADS depression score was associated with an increase in QLACS-GSS of 0.62 (95% CI 326 0.04, 1.23) and 1.86 (95% CI 0.92, 2.79), respectively. Two baseline characteristics were 327 associated with a lower QLACS-GSS at 12 months (better QOL): being over the age of 50 (-328 11.83 95% CI -18.89, 4.78); and a higher self-efficacy score (-2.90 95% CI -4.33, -1.47). Being 329 treated with surgery following cancer diagnosis was also associated with a lower QLACS-GSS 330 at 12 months (better QOL) (-10.21 95% CI -19.09, -1.33). 331 332 Model four: association between baseline characteristics and QLQ-C30 summary score at 12 333 months 334 Baseline characteristics that were associated with a lower QLQ-C30 summary score (worse 335 QOL) at 12 months include having either one (-4.60 95% CI -7.26, -1.94) or two or more (-336 8.54 95% CI -11.41, -5.67) limiting comorbidities and being obese (-2.84 95% CI -5.64, -0.04). 337 A one unit increase in HADS depression score was associated with a decrease in QLQ-C30 338 summary score of 0.98 (95% CI -1,46, -0.50). Living in the most deprived quintile (-4.66 95% 339 CI -8.49, -0.84) and having a lower SES (-6.35 95% CI -9.85, -2.84) were both associated with 340 worse QOL at 12 months. Two baseline characteristics were associated with a higher QLQ-341 C30 summary score (better QOL) at 12 months: a one unit increase in self-efficacy was 342 associated with an increase in QLQ-C30 summary score of 1.81 (95% CI 1.07, 2.55). 343 Receiving surgery as a treatment was also associated with an increase in QLQ-C30 summary 344 score (5.78 95% CI 1.19,10.37). For the cohort as a whole, there was a significant reduction in QOL between baseline and 359 three months. The results at 12 months differed slightly depending on which instrument 360 was used to assess QOL. On the QLACS-GSS, although QOL improved between three and 12 361 months, the difference was not statistically significant and QOL did not return to baseline 362 levels. On the QLQ-C30 summary score, QOL was significantly better at 12 months 363 compared to baseline. Whilst the QLQ-C30 summary score includes some scales which are 364 likely to have long term applicability, it also includes symptoms which are more prevalent 365 during the treatment phase, e.g. nausea and vomiting. The QLACS-GSS was developed 366 specifically to evaluate QOL in longer term survivors and so covers some additional areas, 367 such as sexual functioning. It also frames some items in terms of how bothersome a 368 symptom or issue is rather than whether it is simply present or absent. Although the QLACS-369 GSS and the QLQ-C30 summary score are both useful QOL tools, it may be that the QLACS-370 GSS is more sensitive to longer term QOL issues than the QLQ-C30 summary score. These 371 findings serve as a reminder of how QOL tool selection can influence results, the importance 372 of selecting the right instrument depending on purpose and the benefit of using 373 complementary measures. 374 375 Clinicians may find it helpful to share the QOL trajectories with patients, so they know what, 376 "on average", to expect, including the dip in QOL at three months. Of course, it is important 377 to explain to patients that there are individuals whose QOL does not follow the same 378 pattern. In addition, the regression models can potentially help clinicians identify women at 379 risk of worse QOL around diagnosis and at 12 months. 380 At baseline, factors associated with worse QOL for both outcome measures were limiting 382 comorbidities, anxiety and depression. In addition, having a cancer stage >1 or living in the 383 most deprived areas were associated with low QLQ-C30 summary scores. Women with a 384 better self-reported health status and higher self-efficacy had better QOL on both measures, 385 and unexpectedly, so did women who were obese. Previous research in a mixed 386 gynaecological sample has found that obesity is associated with worse QOL [29]. We also 387 found that being single/separated/divorced was associated with a higher QLQ-C30 summary 388 score, being older than 50 years and being physically active were associated with better QOL 389 as assessed by the QLACS-GSS, consistent with a previous longitudinal study which found 390 that younger women with gynaecological cancer reported lower QOL than older women 391 [30]. Some of these findings may be due to the women with endometrial cancer, who made 392 up about two thirds of the total cohort. They were more likely to be older and obese, and 393 they tended to be diagnosed at an earlier stage. Details are shown in the supplementary 394 material, tables S5 and S6. 395 396 At 12 months, baseline factors associated with worse QOL on both measures were limiting 397 comorbidities, depression and low SES. In addition, being obese or living in the most 398 deprived areas at baseline were associated with worse QOL (lower QLQ-C30 summary 399 scores) at 12 months. A grade 3 tumour was associated with worse QOL as assessed by the 400 QLACS-GSS at 12 months. The different findings for obesity at baseline and 12 months 401 suggest that obese women tolerated treatment less well. An example of the potential 402 impact of obesity is that just 54% of women with endometrial cancer who were morbidly 403 obese had a laparoscopic hysterectomy compared with 63% of those who were not 404 morbidly obese. Obesity has also been associated with worse QOL in previous studies with endometrial cancer survivors [31]. High self-efficacy at baseline was associated with better 406 QOL on both measures at 12 months. Being older than 50 years at baseline was associated 407 with better QOL at 12 months as assessed by the QLACS-GSS. 408

409
While not a baseline characteristic, we included mode of treatment received within six 410 months of baseline in our regression analysis. We found that having surgery was associated 411 with better QOL at 12 months on both QOL measures. We suggest that the small minority of 412 participants who did not receive surgery may have been less fit than those who did receive 413 surgery, and that this may then be reflected in poorer QOL at 12 months. 414

415
Our findings support the hypothesis that self-efficacy is a key protective factor for QOL: it 416 was a significant factor for both measures and at both time points. Women with higher self-417 efficacy may experience the same challenges as women with lower self-efficacy, but they Our findings also support the hypothesis that depression is a key risk factor for poor QOL: 426 poor mental health, as assessed by the HADS, was identified as a factor for putting women 427 at risk for poor QOL at baseline and at 12 months. Comorbidities which limit everyday life 428 were also a key risk factor. The most common reported comorbidity which impacts everyday life was osteoarthritis, followed by depression (as diagnosed by a health professional) and 430 asthma. These conditions are all treatable, as long as they are identified. Our previous 431 research with colorectal cancer patients also found that poor mental health and 432 comorbidities which limit everyday life were associated with poor outcomes [25,26,34]. 433 The findings highlight the need for individually tailored care plans that address all modifiable 434 factors for women with complex needs. In common with other long-term cohort studies, particularly those which collect large 474 amounts of data directly from participants, there was a reduction in participation levels over 475 time. The response rate to questionnaires at 12 months was 59%; this compares to rates of 476 between 37.7% and 95.6% in other gynaecological cancer cohort studies of varying study 477 design [12, 14, 18]. Descriptive analyses were used to compare decliners and consenters 478 and also the differences between those who responded at 12 months and those who did 479 not. Decliners and consenters were similar by cancer type, age and ECOG status but differed 480 by cancer stage, with consenters more likely to have Stage 1 cancers. There was no 481 significant difference between responders at 12 months and non-responders regarding 482 ECOG status. However, compared to non-responders at 12 months, responders were more 483 likely to have endometrial cancer and more likely to be aged 61-70 years. Differences 484 between responders and non-responders were controlled for in the regression analysis. 485 486

Implications for clinical practice 487
We recommend that every woman with a diagnosis of gynaecological cancer is screened for 488 depression, anxiety and self-efficacy to manage the consequences of cancer and its 489 treatment soon after diagnosis and offered appropriate psychological support. An 490 assessment of whether comorbidities limit daily activities should also be carried out and a 491 care plan, tailored to the individual, should be developed and reviewed regularly to support 492 the delivery of personalised care. Extra support may be required for younger women, those 493 with low socioeconomic status and those who live in deprived areas, along with additional 494 effort to promote existing interventions and support. These activities could form part of a 495 holistic needs assessment and could be included in a prehabilitation programme. 496 497

Conclusion 498
The results from this unique study have identified poor mental health and comorbidities 499 which limit everyday life as key risk factors for poor QOL at diagnosis and 12 months later. 500 The key protective factor is self-efficacy. 501