Background
Cervical cancer (CC) has become one of the most susceptible and lethal tumors for women due to the increase of sexually transmitted diseases. GLOBAOCAN report showed that in 2020, there would be 600,000 new cases of CC worldwide, and the number of deaths due to CC would reach 340,000 [
1]. CC ranked the fourth in the number of new cases of women in the world [
2]. CC has become an important public health problem. At present, surgery and chemotherapy are the most commonly used treatment methods for CC, which can improve the overall survival rate and prolong the life expectancy of patients, but it is difficult to avoid the harm caused by surgical trauma, complications and side effects of radiotherapy and chemotherapy.
Cancer-related fatigue (CRF) often runs through all stages of radiotherapy, chemotherapy and even hospice care for cancer patients [
3]. Piper first proposed the concept of CRF in 1987, defining it as a subjective, specific and systematic feeling of excessive fatigue, which was closely related to the cancer itself and its therapeutic factors [
4]. Ma showed that the overall incidence rate of CRF among 144,813 cancer patients was 52%, and the number of patients with moderate fatigue was significantly higher than that of mild and severe fatigue [
5]. Cancer survivors reported that CRF was a serious and destructive symptom that can last for months to years after treatment [
6]. Gernier et al. followed up 45 patients with CC and found that the proportion of physical fatigue and mental fatigue was 45.2% and 37.8% respectively [
7]. Al Maqbali et al. found that the incidence rate of CRF during treatment and within three months after treatment was as high as 62.0% and 50.1%, respectively, and 43% of the survivors still had fatigue symptoms of varying degrees [
8]. This demonstrated that CRF could occur at various stages of cancer treatment. Research has shown that 60–90% of cancer patients who received treatment experience symptoms of CRF, including physical weakness, silence, and functional impairment [
9]. Compared with patients without CRF, patients with CRF had relatively poorer quality of life, more prominent symptoms of depression and anxiety, and severe physical and cognitive dysfunction [
10]. Overall, severe CRF could affect daily activities [
11], and lead to depressive symptoms [
12], poor quality of life, lack of vitality, work difficulties, relationship issues, emotional distress [
13], and even affect therapeutic compliance and clinical outcomes including recurrence and mortality [
14].
Clinical prediction models (CPMs) are used to evaluate the probability of a specific subject suffering from a certain disease or having a certain clinical result in the future [
15]. CPMs calculate the probability of a certain disease or complication in the current state according to the patient’s clinical symptoms, disease characteristics and other relevant data information [
16]. The prediction model of CRF constructed by Meglio et al. found that age, BMI, current smoking behavior, anxiety, insomnia, and pain during diagnosis were predictive factors, and the accuracy of the model was very high [
17]. Lee et al. also constructed a random forest regression model for CRF in patients with breast cancer, and found a subset of genes with more predictive significance, such as peroxygenase-5, connector protein, and the accuracy of the model was high [
18]. Huang et al. constructed a back-propagation artificial neural network model to predict the risk of moderate to severe CRF in colorectal cancer patients and found surgery, complications, hypokalaemia, albumin, neutrophil percentage, pain, sleep quality, anxiety, depression and nutrition were predictive factors [
19].
As a type of CPMs, a nomogram have been widely used as a prediction method in oncology in recent years [
20‐
23]. It meets the requirements of integrated models, plays a role in promoting personalized healthcare, and is convenient for clinical doctors to use for prognosis prediction [
24]. A nomogram refers to a quantitative analysis chart that represents the functional relationship between multiple variables using a set of disjoint line segments in planar coordinates [
24]. Its advantage is that it can directly use the graph to calculate the value of a certain variable, such as the patient’s indicator score or survival probability [
24]. The most common one is the probability nomogram, which determines the probability of a specific event occurring in an individual, such as disease occurrence, recurrence, and prognosis (such as death) [
25]. Essentially, a nomogram is a visualization of the results of a regression equation, commonly used for displaying the results of logistic regression or COX regression [
26]. Based on the regression results, multiple line segments are drawn in specific proportions, and through plotting, the disease risk or survival probability of an individual can be conveniently calculated [
26]. Many studies have used a nomogram to predict the probability of fatigue occurrence in different populations, and have validated the accuracy of the nomogram [
27‐
30].
However, a nomogram of sever CRF in patients with CC was rarely reported. Therefore, we included the factors that have been confirmed by previous studies that might affect CRF, including age, economic status, exercise status, clinical status and psychological variables [
27‐
32]. This study aimed to develop and validate a scientific, accurate and convenient new assessment tool for the prediction of severe CRF in patients with CC, so as to help clinical workers identify high-risk groups with severe CRF in CC as early as possible.
Discussion
This study showed 283 patients with CC had CRF of different degrees, and the incidence was as high as 99%, of which the incidence of mild and moderate CRF was 53.2%, and the incidence of severe CRF was 46.8%. Through the nomogram above, we learned that in addition to demographic and clinical characteristics, patients’ psychological conditions were more influential, similar to the model of CRF in patients with breast cancer [
17].
Risk factors of sever CRF in patients with CC
Our study found that long-term passive smoking was risk factor of sever CRF. The reason may be that many carcinogenic and toxic chemicals in second-hand smoke have high concentrations, leading to malignant diseases [
46], or passive smoking patients have more negative emotions and poorer sleep disorders, which may exacerbate CRF [
47]. We found that tumor recurrence was risk factor of sever CRF. It may be due to patients with tumor recurrence feeling fearful of the disease, suspecting the possibility of curing the disease, affecting their confidence in treatment, and having poor mental health, which in turn exacerbates CRF [
48]. Our study also found that negative coping style (avoid or yield) was risk factors for sever CRF. Perhaps it is because negative coping style can affect the recovery process of cancer patients, leading to a cold and negative attitude towards their own diseases. Over time, this can increase the psychological burden on patients and lead to CRF [
49].
Protective factors of sever CRF in patients with CC
Our study found that patients with monthly income > 5000 CNY had a lower risk of severe CRF. Perhaps it is because patients usually face high medical costs after diagnosis, which brings greater psychological pressure to low-income patients [
17], and may lead high-risk CRF. We found that patients who exercised ≥ 2–3 times a week had a lower risk of severe CRF. This is because exercise can increase the body’s blood oxygen content, accelerate metabolism, stimulate the central nervous system, and improve the patients’ mental state, thereby eliminating CRF [
50]. Our study found that patients who experienced higher social support have relatively lower CRF, which may be due to the social support provided by role relationships helping to stabilize and develop positive self-esteem and confidence, enhancing patients’ ability to withstand stress, and reducing CRF [
51]. We also found that patients with higher SOC had a lower risk of developing severe CRF. This is because there are physiological and psychological stressors in the diagnosis and treatment of cancer, and SOC can strengthen the management of corresponding stressors, enabling patients to maintain good physical and mental health outcomes [
52].
Evaluation and analysis of the nomogram
The areas under the ROC curve of both groups were greater than 0.8, indicating that the nomogram can better distinguish severe CRF patients [
45]. In the consistency test, the calibration curves were well fitted (
P < 0.05) in both groups, indicating that the probability of severe CRF predicted by the nomogram was consistent with the actual probability of severe CRF in patients, and the accuracy of prediction was high. The DCA analysis showed that the net benefit of applying the nomogram to most thresholds in both groups was good. According to the best cut-off value 0.444 in ROC curve, patients with CC can be divided into high-risk group and low-risk group of CRF. In addition, this study visualized the the regression equation results in the form of the nomogram, which was more intuitive and convenient for calculation, and was conducive to the practical application of the model in clinical practice [
53]. For patients whose prediction probability was close to or higher than the optimal threshold, early intervention could be carried out according to their coping style, social support, SOC and so on.
Clinical implications
Our study has developed the first nomogram of CRF for patients with CC. It can strengthen the risk identification of severe CRF, and its independent risk factors provided scientific basis for patients to implement intervention measures. For example, if a patient exercises ≥ 2–3 times/week, has a per capita income of > 5000CNY, and has high social support characteristics, their scores for exercise, income, and social support can be calculated based on the nomogram. Then, the above scores are added up to obtain the total score of the patient. Based on the nomogram, estimate the probability of sever CRF occurrence corresponding to the total score, that is, the probability of patient experiencing sever CRF. This nomogram was significant for strengthening risk management, reducing or controlling the occurrence of severe CRF.
Limitations
The nomogram developed in this study may have the following limitations. Firstly, the predicted results of the nomogram remain unchanged over time, but in fact, the outcomes of disease are changing with improvements in treatment, early detection, and changes in natural history, therefore, over time, the performance of the nomogram may become less accurate. Secondly, although studies have shown that nomogram is superior to the judgment of clinical doctors, however, the conclusion is purely based on AUC and does not equate to improving clinical efficacy. Again, although nomogram can be used to define the effectiveness of clinical trials, treatment decisions for these cases should follow the inclusion criteria determined by the nomogram and the subsequent benefits related to treatment, rather than just the estimated risks in the nomogram. Finally, although the nomogram performs well, the evaluation of whether it can improve patient and doctor satisfaction, and tumor prognosis is often overlooked.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.