Access to this full-text is provided by Springer Nature.
Content available from Scientific Reports
This content is subject to copyright. Terms and conditions apply.
Construction and validation of
prognostic models for young
cervical cancer patients: age
stratication based on restricted
cubic splines
Yuan Gong1,2, Feifei Gou1,2, Qingfeng Qin1, Weijie Tian1, Wei Zhao1 & Dan Zi1
Cervical cancer (CC) ranks as the second highest cause of morbidity and mortality among young
women; however, there are currently no age-specic denitions for young cervical cancer or prognostic
models tailored to this demographic. Data on CC diagnosed between 2000 and 2019 were extracted
from the Surveillance, Epidemiology, and End Results (SEER) database. Age stratication is based on
the relationship between age and cancer-specic mortality, as demonstrated by restricted cubic spline
analyses (RCS). Cox proportional hazards regression analyses were employed to identify independent
prognostic factors in the young CC group. Two novel nomograms for this population were developed
and validated using an external validation cohort obtained from a local hospital database, evaluated
with concordance index (C-index) and calibration plots. Receiver operating characteristic (ROC) curves
were utilized to compare the accuracy of the established models against the International Federation
of Gynaecology and Obstetrics (FIGO) staging system (2018). A total of 27,658 patients from the
SEER database were classied into three age groups (<36 years, 36-60 years, >60 years) based on
RCS analyses, with 4,990, 16,922, and 5,746 patients in each group, respectively. The independent
prognostic factors identied for young CC included stage, tumour size, grade, histologic type, and
surgical intervention. The results of the C-index and calibration in both the training and validation sets
conrmed that the two nomograms can accurately predict the occurrence and prognosis of young CC
patients. The area under the curve (AUC) values indicated that these models demonstrated higher
ecacy in predicting overall survival (OS) compared to the FIGO staging system (2018). These models
could potentially serve as eective tools for clinicians to estimate the prognosis of young CC patients.
Cervical cancer (CC) is the fourth most frequently diagnosed cancer and the fourth leading cause of cancer death
in women, with an estimated 604,000 new cases and 342,000 deaths worldwide in 2020, which represents a major
global health challenge1. Boosting rates of HPV vaccination and eective screening have reduced the incidence
of CC, and the World Health Organization (WHO) is committed to eradicating cervical cancer by achieving an
incidence rate of no more than 4 cases per 100 000 women-years worldwide2. However, in adolescents and young
women, CC remains the malignancy with the second highest morbidity and fatality rate3,4; moreover, increases
in the incidence or mortality of CC among young women have been reported in some regions, such as urban
China, Japan, Eastern European countries and Latin America5–10.
Adolescents and young women represent a distinct demographic in cancer research. Previous studies11–14 on
young cervical cancer have encompassed a range of ages from 25 to 45 years, leading to a lack of consensus on
the denition of young cervical cancer. Additionally, previous studies posited that young cervical cancer were
involved in more aggressive pathogenesis, resulting in a poorer prognosis15–17; thus, youth was recognized as a
prognostic factor for cervical cancer. However, to the best of our knowledge, there is currently a limited number
of studies providing reliable data on the clinicopathological features and prognostic factors associated with
young CC, and no predictive model for its prognosis has been established to date. In this study, we performed
age stratication by a novel method and conducted this retrospective study to analyse the clinicopathological
characteristics, treatments, and prognosis of young cervical cancer patients. We established visual prognostic
1Department of Gynaecology, Guizhou Provincial People’s Hospital, Guiyang550001, China. 2These authors
contributed equally: Yuan Gong and Feifei Gou. email: zidangy08@163.com
OPEN
Scientic Reports | (2024) 14:29808 1
| https://doi.org/10.1038/s41598-024-81644-z
www.nature.com/scientificreports
Content courtesy of Springer Nature, terms of use apply. Rights reserved
models as a supplement to International Federation of Gynaecology and Obstetrics (FIGO) staging to better
predict the prognosis of young CC patients, which has not been studied before.
Methods
Database and participants
e Surveillance, Epidemiology, and End Results (SEER) database (https://seer.cancer.gov/) is a US p o p u l a t i o
n - b a s e d cancer registry database. We used SEER*Stat soware (version 8.4.1) to extract information on patients
diagnosed with cervical cancer between 2000 and 2019. e primary topographic site was selected using ICD
codes for Endocervix (C53.0), Exocervix (C53.1), Overlapping lesions of cervix uteri (C53.8) and Cervix uteri
(C53.9). e data used for external validation were obtained from the Gynaecological database of the Guizhou
Provincial People’s Hospital, and information on patients with cervical cancer diagnosed between 2014 and 2020
was extracted. Inclusion was based on the following: rst diagnosis of cervical cancer with primary focus and
diagnosis conrmed by histopathology. e exclusion criteria were as follows: (1) survival time < 1 month and
(2) missing data for the chosen variables.
Variables
Risk factors used for the analysis included age, marital status at diagnosis, median household income (ination-
adjusted), months from diagnosis to treatment, grade, histologic type (squamous, adenocarcinomas, and others
including uncommon subtypes such as complex epithelial neoplasms, small cell carcinoma and adenosquamous
cell carcinoma), tumour size (< 2cm, 2–4 cm, > 4cm), stage, surgery, radiotherapy and chemotherapy. Since
2018, the pathological grading of tumours reported in the SEER database has been in accordance with a
3-category classication system, meaning that grade 4 (undierentiated; anaplastic) was combined with grade 3
(poorly dierentiated). To ease the interpretation of the results, we converted all data before 2018 into the 3-stage
system. Besides, the FIGO staging system of cervical cancer was changed aer 2018, mainly to include stage III
lymph node metastasis. To reect the role of lymph node metastasis in the prognosis of cervical cancer and be
convenient for gynaecological clinical application, cases included were converted to 2018 FIGO stage (I, II, III,
IV) in our study according to the SEER TNM staging data [AJCC 3rd (1988–2003) TNM data; Derived AJCC
TNM stage, 6th edition (2004–2015); Derived SEER Combined TNM stage (2016–2017); Derived EOD 2018
TNM stage (2018+)]. In addition, according to the reported median annual household income in the United
States in a previous study, $60,000 was set as the cut-o value to group the patients18.
Outcomes of interest
Overall survival (OS) and cancer-specic survival (CSS) were set as outcomes. OS was dened as the period
from the date of diagnosis until death attributed to any cause or to the end of follow-up. CSS was dened as the
interval from the date of diagnosis until death as a result of cervical cancer or to the end of follow-up.
Statistical analysis
Continuous data are presented as medians and ranges, and categorical data are shown as frequencies and
proportions. e Kaplan‒Meier method in X-tile soware (Version 3.6.1, http://tissuearray.org) was used to
evaluate the optimal cut-o value for months from diagnosis to treatment (Supplement 1). e relationship
of age and the cancer-specic mortality of CC was t by univariate Cox regression with restricted cubic spline
(RCS) analyses (knot = 5)19. e cut-o values were set for age stratication in CC patients based on the inection
points in RCS. OS and CSS of dierent age groups were estimated using the Kaplan‒Meier method. Univariate
Cox regression analysis was used to calculate the risk ratio (hazard ratio, HR) and the associated 95% condence
interval (CI) to screen potential prognostic factors aecting the OS and CSS of young CC patients from the SEER
database; variables with statistically signicant dierences (p value < 0.05) were included in the multivariate Cox
regression model to screen for independent risk factors (p value < 0.01) for prognosis in young CC patients.
A nomogram model was constructed based on the results of the multivariate Cox regression analysis and
assessed using the external validation cohort. e predictive ability of the model for death was assessed by the
concordance index (C-index), and a higher C-index indicates a better ability to separate patients with dierent
survival outcomes. e slopes of the calibration curves were used to compare the predicted probability with
the observed probability in the study cohort. e area under the receiver operating characteristic (ROC) curve
(AUC) was used to compare the dierence in the prediction accuracy between the model and the FIGO staging
system (2018). All analyses were performed using R soware, version 4.2.2 (R Project for Statistical Computing).
Ethical statement
We obtained signed authorization and permission to access and use the data from the SEER database and
followed the protocol throughout the process to protect patient privacy. Furthermore, we retrospectively
collected data from medical record system of Guizhou Provincial People’s Hospital, and all patients’ personal
information was anonymized, so the ethical review board of Guizhou Provincial People’s Hospital approved
the ethical requirements for this study and waived the requirement of obtaining written informed consent from
patients. is study was conducted in accordance with the revised Declaration of Helsinki.
Result
Age-stratied incidence and outcomes of CC
A total of 27,658 cases of cervical cancer from the SEER database were rst enrolled in our study based on the
inclusion and exclusion criteria (Supplement 2). e RCS presented a nonlinear relationship between age and
cervical cancer-specic mortality (p < 0.001) (Fig.1). e plot showed a reduction trend of risk within the lower
Scientic Reports | (2024) 14:29808 2
| https://doi.org/10.1038/s41598-024-81644-z
www.nature.com/scientificreports/
Content courtesy of Springer Nature, terms of use apply. Rights reserved
age range; the lowest risk was reached at age 35.58 years old [HR = 0.687, 95% CI (0.631, 0.749)], and then risk
increased thereaer with a substantial increase aer 60 years old. Based on these results, 36 and 60 years were
set as the cut-o values for age stratication in our study. e KM curves (Fig.2) present the OS and CSS of
cervical cancer patients in three dierent age groups (< 36, 36–60, > 60). Patients in the young group had the
best prognosis, followed by those in the middle-aged group, and the 5-year OS and CSS in both groups were
over 70%. Undoubtedly, patients in the elderly group had the worst prognosis (P < 0.001), with a 5-year OS rate
of approximately 50%. Aer 5 years of cervical cancer diagnosis and treatment, the CSS KM curve of all groups
tended to be at, which also veried the regularity of disease progression and conrmed the necessity of a
standardized follow-up visit.
Baseline characteristics of young group patients and dierences from the other two groups
4,990 cervical cancer patients (18.0%) younger than 36 years old were assigned to young group. In terms of
demographic characteristics, nearly half of the young patients were never married (44.4%), and the proportion
of patients with a relatively lower median household income (<$60,000) in this group was higher than that in
the other groups (29.9% vs. 28.7% vs. 27.3%). ose patients exhibited a high compliance with treatment aer
diagnosis, with a higher proportion accepting treatment within 1 month (42.6% vs. 38.6% vs. 31.6%).
Regarding the clinical characteristics, in the young group, the proportion of other histologic types
(nonsquamous cell neoplasms and nonadenocarcinoma) was higher than that in the other two groups (8.3% vs.
7.2% vs. 7.5%). e percentages of early-stage (FIGO stage I), grade 1 and smaller tumour sizes (< 2cm) were
higher than those of the other two groups (62.3% vs. 52.3% vs. 38.6%, 16.6%vs. 14.9%vs. 9.4%, 39.3% vs. 29.8%
vs. 20.1%, respectively). e demographic and clinical characteristics of the dierent groups are summarized in
Table1.
Fig. 1. e relationship of age and the cancer specic mortality of CC t by univariate Cox regression
with RCS analyses. e Cox regression indicated a nonlinear relationship between age and the risk of
death(p < 0.001), with the lowest risk of mortality at age = 35.58 years. When the age was older than 35.58
years, HR was positively correlated with age, whereas negative correlations between age and HR were identied
when age was less than 35.58 years. When age was older than 60 years, the risk of death increased substantial
with increasing age. Shaded areas represent 95% CI.
Scientic Reports | (2024) 14:29808 3
| https://doi.org/10.1038/s41598-024-81644-z
www.nature.com/scientificreports/
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Prognostic factors for young cervical cancer patients
e young age group was used as a training cohort for developing prognostic models for young, early-onset
cervical cancer. Univariate and multivariate Cox regression analyses were successively performed to identify
signicant risk factors in the training cohort (Table2; Fig.3). Pathologic type, FIGO stage, tumour size and
surgery were included in the nal regression models. According to the multivariate Cox regression analysis
results of OS, the grade did not meet the qualication of p < 0.01, but as a recognized prognostic factor, it was
also incorporated into the nal model. In addition, chemotherapy and radiotherapy status were signicant high-
risk factors with p < 0.01 but were not included in either model because patients who received chemotherapy
or radiotherapy were also worse in tumour stage or grade, which might lead to logical errors or issues of
multicollinearity in the models.
Visualization and validation of the prognostic models
Nomograms for OS and CSS respectively were presented for visualization of the prognostic models (Fig.4).
Apparently, stage had the greatest eect on prognosis, followed by tumour size. Contrary to common
understanding, squamous carcinoma was associated with a worse prognosis than adenocarcinoma, and both
were better than other types (nonsquamous cell neoplasm and nonadeocarcinoma). e C-indexes of the OS
and CSS prognosis were 0.805 (95% CI: 0.793–0.817) and 0.820 (95% CI: 0.808–0.832), respectively.
A total of 125 patients from the gynaecological database of Guizhou Provincial People’s Hospital were
included as the validation cohort in our study based on the inclusion and exclusion criteria. e demographic
and clinical characteristics of this cohort were presented in Table S1. In the validation cohort, since all patients
died from cervical cancer, we studied OS only, and the C-index was 0.865 (95% CI: 0.800–0.930).
In addition, the calibration plots of the training cohort and validation cohort showed a high t between
the nomogram-predicted survival and actual survival in terms of the probability of 3- and 5-year OS and CSS
(Fig.5).
Comparison of the nomogram with the FIGO staging system
e ROC curve was used to compare the accuracy between the established model and the FIGO staging system
(2018). We plotted the ROC curves for the 3- and 5-year OS in the external validation cohort and calculated
the AUC values (Fig.6). e results showed that our model had higher ecacy in predicting OS than the FIGO
staging system (2018).
Discussion
At present, the prognostic models of cervical cancer are gradually rened into special pathological subtypes,
stage or disease status (such as lymph node metastasis) and special populations, and the accuracy of the models
is excellent11,20,21. A prognostic model for young, early-onset cervical cancer has not been reported. e question
is how to dene ‘young’ patients. Previous reports have studied patients aged 25 years or younger as a very young
group because the clinicopathological characteristics and pathogenesis of this population may be dierent from
those of adult women over 25 years old12,13. ere are also studies setting younger patients as under 30, 35 or
40 years with no explanation14. e results regarding the relationship between age and mortality in cervical
cancer were inconsistent in studies with dierent samples, which may also point to a special relationship
between age and prognosis. In this study, we applied a novel method to dene the young group and performed
age stratication for cervical cancer. RCS models have been proven to be a good tool to t the nonlinear
association between continuous variables and outcomes, and it has been conrmed that RCS, together with the
Cox proportional hazards regression model, enables visualization of the shapes of dose‒response associations
between a continuous variable and outcome risks22,23. e RCS plot presented a nonlinear relationship between
age and mortality of CC patients in our study. Age of 36 and 60 were the inection points in the plot; thus, it was
Fig. 2. Kaplan-Meier curves of OS (a) and CSS (b) for 3 groups of CC patients.
Scientic Reports | (2024) 14:29808 4
| https://doi.org/10.1038/s41598-024-81644-z
www.nature.com/scientificreports/
Content courtesy of Springer Nature, terms of use apply. Rights reserved
reasonable to divide patients into 3 age groups based on these two cut-o values, and patients less than 36 years
old were assigned to the young group.
Previous studies suggested that younger age was associated with a worse prognosis16,17. However, our
ndings indicate that the younger group exhibited the best prognosis, characterized by a higher proportion of
early tumor stages, smaller tumor sizes, grade 1 tumors, and better medical compliance. Some other studies have
reached similar conclusions12,14,24,25. Regarding prognostic factors, histologic type is widely acknowledged as
signicant in tumours. Ruiz et al. suggested that histology substantially aected OS of CC24. Certain pathological
subtypes of CC exhibit more aggressive behavior and are not associated with human papillomavirus (HPV)
infection. HPV-negative CC was considered to be associated with a poorer prognosis26,27, and the sensitivity
of cytological screening for these subtypes was lower than that for cervical squamous cell carcinoma (CSCC)28,
potentially resulting in delays in diagnosis. In our study, the proportions of nonsquamous cell neoplasms and
nonadenocarcinoma types were higher in the young group but still less than 10% (8.4%), which may have
had little impact on the overall prognosis. Similarly, in other studies of cervical cancer in young patients, no
signicant relation was found between histological type and survival12,25. In addition, the increasing incidence
Subject
< 36 36 ~ 60 > 60 overall
(N = 4990) (N = 16922) (N = 5746) (N = 27658)
Age
Median [Min, Max] 31.0 [14.0, 35.0] 46.0 [36.0, 60.0] 68.0 [61.0, 85.0] 47.0 [14.0, 85.0]
Marital status
Married 2295 (46.0%) 8686 (51.3%) 2147 (37.4%) 13,128 (47.5%)
Single(never married) 2215 (44.4%) 4885 (28.9%) 910 (15.8%) 8010 (29.0%)
Other 480 (9.6%) 3351 (19.8%) 2689 (46.8%) 6520 (23.6%)
Median household income
<$60,000 1494 (29.9%) 4850 (28.7%) 1570 (27.3%) 7914 (28.6%)
>=$60,000 3496 (70.1%) 12,072 (71.3%) 4176 (72.7%) 19,744 (71.4%)
Histologic subtype
Squamous cell neoplasms 3308 (66.3%) 10,945 (64.7%) 3950 (68.7%) 18,203 (65.8%)
Adenocarcinoma 1269 (25.4%) 4762 (28.1%) 1366 (23.8%) 7397 (26.7%)
Other 413 (8.3%) 1215 (7.2%) 430 (7.5%) 2058 (7.4%)
Tumour size (cm)
< 2 1961 (39.3%) 5049 (29.8%) 1154 (20.1%) 8164 (29.5%)
2 ~ 4 1437 (28.8%) 4959 (29.3%) 1919 (33.4%) 8315 (30.1%)
> 4 1592 (31.9%) 6914 (40.9%) 2673 (46.5%) 11,179 (40.4%)
Surgery
No 1101 (22.1%) 5380 (31.8%) 2662 (46.3%) 9143 (33.1%)
YES 3889 (77.9%) 11,542 (68.2%) 3084 (53.7%) 18,515 (66.9%)
Stage
I 3108 (62.3%) 8854 (52.3%) 2220 (38.6%) 14,182 (51.3%)
II 461 (9.2%) 2401 (14.2%) 1230 (21.4%) 4092 (14.8%)
III 1148 (23.0%) 3988 (23.6%) 1442 (25.1%) 6578 (23.8%)
IV 273 (5.5%) 1679 (9.9%) 854 (14.9%) 2806 (10.1%)
Months from diagnosis to treatment
< 1 2124 (42.6%) 6531 (38.6%) 1815 (31.6%) 10,470 (37.9%)
>=12866 (57.4%) 10,391 (61.4%) 3931 (68.4%) 17,188 (62.1%)
Radiation
Beam radiation 1181 (23.7%) 4920 (29.1%) 2145 (37.3%) 8246 (29.8%)
Combination 1064 (21.3%) 4616 (27.3%) 1734 (30.2%) 7414 (26.8%)
Implants 149 (3.0%) 659 (3.9%) 280 (4.9%) 1088 (3.9%)
None/Unknown 2596 (52.0%) 6727 (39.8%) 1587 (27.6%) 10,910 (39.4%)
Grade
Grade1 826 (16.6%) 2529 (14.9%) 541 (9.4%) 3896 (14.1%)
Grade2 2134 (42.8%) 7476 (44.2%) 2376 (41.4%) 11,986 (43.3%)
Grade3 2030 (40.7%) 6917 (40.9%) 2829 (49.2%) 11,776 (42.6%)
Chemotherapy
No/Unknown 2830 (56.7%) 8077 (47.7%) 2558 (44.5%) 13,465 (48.7%)
Yes 2160 (43.3%) 8845 (52.3%) 3188 (55.5%) 14,193 (51.3%)
Tab le 1. e demographic and clinical characteristics of 3 groups.
Scientic Reports | (2024) 14:29808 5
| https://doi.org/10.1038/s41598-024-81644-z
www.nature.com/scientificreports/
Content courtesy of Springer Nature, terms of use apply. Rights reserved
of young cervical cancer patients may be related to social factors. In this study, the proportion of single (never
married) people in the young group was signicantly higher than that in the other two groups, with a narrow gap
in median household income compared to the middle age group. Being single and more likely to have multiple
sexual partners29, promotes persistent HPV infection.
e prognostic model illustrated by the visualized nomogram has been employed in clinical research across
various tumours30–32, including studies on cervical cancer33,34. In this study, nomograms were utilized to develop
a prognostic model specically for young cervical cancer patients. e independent prognostic factors identied
included stage, tumour size, grade, histologic type, and surgical intervention, consistent with the ndings of
previous studies35,36. e relative weight of each prognostic factor was visually represented in the nomograms,
demonstrating that each young cervical cancer patient with available clinical data on these prognostic factors
could derive a specic score from the nomograms. is score could subsequently be utilized to predict 3- year
or 5- year survival outcomes. Evaluation metrics for model accuracy included the C-index and the calibration
curve, among others. We utilized the SEER database and incorporated statistically signicant prognostic factors
Subject
Cancer-specic survival(CSS) Overall sur vival(OS)
HR(95%CI) p-value HR(95%CI) p-value
Age(years)
< 26 Reference Reference
26 ~ 35 0.732 ( 0.588 ~ 0.910) 0.005** 0.777(0.633 ~ 0.954) 0.016*
Median household income
<$60,000 Reference Reference
>=$60,000 0.813(0.705 ~ 0.939) 0.005** 0.811(0.712 ~ 0.924) 0.002**
Marital status
Married Reference Reference
Other 1.267(0.997 ~ 1.611) 0.053 1.292(1.039 ~ 1.605) 0.021*
Single(never married) 1.517(1.312 ~ 1.753) < 0.001*** 1.557(1.365 ~ 1.777) < 0.001***
Months from diagnosis to treatment
< 1 Reference Reference
≥ 1 1.672(1.444 ~ 1.934) < 0.001*** 1.636(1.433 ~ 1.867) < 0.001***
Histologic type
Adenocarcinoma Reference Reference
Other 3.428(2.665 ~ 4.408) < 0.001*** 2.936(2.336 ~ 3.691) < 0.001***
Squamous cell neoplasms 2.034(1.674 ~ 2.472) < 0.001*** 1.873(1.578 ~ 2.224) < 0.001***
Tumour size (cm)
< 2 Reference Reference
2 ~ 4 4.418(3.362 ~ 5.807) < 0.001*** 3.591(2.850 ~ 4.525) < 0.001***
> 4 13.678(10.630 ~ 17.599) < 0.001*** 10.748(8.706 ~ 13.269) < 0.001***
FIGO stage
I Reference Reference
II 4.053(3.209 ~ 5.119) < 0.001*** 3.689(3.005 ~ 4.529) < 0.001***
III 5.169(4.355 ~ 6.135) < 0.001*** 4.397(3.777 ~ 5.119) < 0.001***
IV 18.464(15.104 ~ 22.573) < 0.001*** 14.926(12.427 ~ 17.926) < 0.001***
Grade
1 Reference Reference
2 2.892(2.094 ~ 3.993) < 0.001*** 2.235(1.723 ~ 2.899) < 0.001***
3 5.166(3.769 ~ 7.081) < 0.001*** 3.768(2.925 ~ 4.853) < 0.001***
Surgery
NO Reference Reference
YES 0.208(0.182 ~ 0.238) < 0.001*** 0.227(0.200 ~ 0.257) < 0.001***
Chemotherapy
No/Unknown Reference Reference
YES 6.281(5.310 ~ 7.431) < 0.001*** 5.341(4.615 ~ 6.181) < 0.001***
Radiotherapy
Beam radiation Reference Reference
Combination 0.837(0.715 ~ 0.979) 0.026* 0.831(0.719 ~ 0.961) 0.016*
Implants 0.957(0.704 ~ 1.300) 0.778 0.944(0.709 ~ 1.258) 0.691
None/Unknown 0.154(0.127 ~ 0.186) < 0.001*** 0.187(0.158 ~ 0.220) < 0.001***
Tab le 2. Univariate Cox regression analyzing the risk factors for young cervical cancer patients.
Scientic Reports | (2024) 14:29808 6
| https://doi.org/10.1038/s41598-024-81644-z
www.nature.com/scientificreports/
Content courtesy of Springer Nature, terms of use apply. Rights reserved
to develop a clinical prognostic model specically for young cervical cancer patients, which has not been
previously reported. e C-index for the OS model and the CSS model for young cervical cancer patients was
0.805 (95% CI: 0.793–0.817) and 0.820 (95% CI: 0.808–0.832), respectively. ese values are higher than those
reported for cervical cancer prognostic models in other studies37,38. In the validation set, the C-index reached
0.865 (95% CI: 0.800–0.930), conrming the good predictive ability of the nomograms. e calibration plots
showed a high t between the nomogram-predicted survival and actual survival, also providing proof for the
robust clinical value of the nomograms.
e FIGO staging system has been updated to the 2018 version, which has been shown to provide better
predictions of disease-free survival and OS compared to earlier FIGO staging versions. In previous prognostic
models, stage was typically the most signicant risk factor for malignant tumours11,39,40. Similarly, the FIGO
stage exhibited the greatest impact on prognosis in our models. However, Grigsby et al. suggested that clinical
outcomes remained heterogeneous despite having the same FIGO stage41. is observation indicated inherent
limitations within the FIGO staging system. It primarily considers the extent, size, and metastasis of the tumor,
Fig. 4. Nomograms for predicting OS (a) and CSS (b) of young CC patients.
Fig. 3. Forest plots for OS (a) and CSS (b) of young CC patients based on multivariate Cox regression analysis
of the training cohort.
Scientic Reports | (2024) 14:29808 7
| https://doi.org/10.1038/s41598-024-81644-z
www.nature.com/scientificreports/
Content courtesy of Springer Nature, terms of use apply. Rights reserved
while neglecting other prognostic factors such as pathological type, grade, and clinical treatment42. ese factors
were incorporated into our prognostic models, thereby enhancing the accuracy of the predictions, as evidenced
by the superior AUC values for our models compared to those for the FIGO staging. Furthermore, the data for
these factors are readily accessible in clinical practice.
ere were some drawbacks that must be recognized in this study. Firstly, e SEER database did not contain
detailed data on clinical characteristics, such as lymphovascular space invasion and HPV subtype, that are crucial
for assessing the prognosis of CC. Secondly, the lack of information on lifestyle, such as age at rst sexual activity,
number of sexual partners and smoking history, limited the comprehensiveness of the study. Furthermore, the
validation cohort was derived from a single center with a relatively small sample size (N = 125), which may raise
Fig. 5. Calibration plots of the models for predicting 3-, and 5-year OS and CSS of the development cohort
(a–d) and 3-, and 5-year OS of external validation cohort (e, f).
Scientic Reports | (2024) 14:29808 8
| https://doi.org/10.1038/s41598-024-81644-z
www.nature.com/scientificreports/
Content courtesy of Springer Nature, terms of use apply. Rights reserved
concerns regarding potential bias. erefore, further studies are necessary to improve accuracy and promote the
model’s adoption.
Conclusion
Based on the established relationship between age and mortality, we dened young CC as occurring in
individuals under the age of 36. e young CC group exhibited a better prognosis compared to other age groups,
with independent prognostic factors including disease stage, tumor size, grade, histologic type, and surgical
intervention. Recognizing that young CC patients represent a distinct cohort with unique clinicopathological
features, we developed and validated prognostic models specically tailored for this population. ese models
demonstrated good calibration and high accuracy, potentially serving as a valuable supplement to the current
FIGO staging system.
Data availability
e data used or analyzed during the current study are available from the corresponding author on reasonable
request.
Received: 28 December 2023; Accepted: 28 November 2024
References
1. Sung, H. et al. Global Cancer Statistics 2020: Globocan estimates of incidence and mortality worldwide for 36 cancers in 185
countries. CA Cancer J. Clin. 71 (2021).
2. Global Strategy to Accelerate the Elimination of Cervical Cancer as a Public Health Problem. (2023). h t t p s : / / w w w . w h o . i n t / p u b l i c a t i o
n s / i / i t e m / 9 7 8 9 2 4 0 0 1 4 1 0 7 (accessed 30 July 2023).
3. You, L. et al. Worldwide Cancer statistics of adolescents and young adults in 2019: A systematic analysis of the global burden of
Disease Study 2019. ESMO Open 6, 27–52 (2021).
4. Aida, Y. et al. Characteristics of cancers in adolescents and young adults compared with those in adults in their 60S: A single-center
experience. Oncology 100, 140–147 (2022).
5. Chen, T. et al. Rising mortality rate of cervical cancer in younger women in Urban China. J. Gen. Intern. Med. 35, 593 (2020).
6. Motoki, Y. et al. Increasing trends in cervical cancer mortality among young Japanese women below the age of 50 years: An analysis
using the Kanagawa population-based cancer registry, 1975–2012. Cancer Epidemiol. 39, 700–706 (2015).
7. Torres-Roman, J. S. et al. Cervical Cancer mortality among Young women in Latin America and the Caribbean: Trend Analysis
from 1997 to 2030. BMC Public Health 22, 113 (2022).
8. Yuan, M., Zhao, X., Wang, H., Hu, S. & Zhao, F. Trend in cervical cancer incidence and mortality rates in China, 2006–2030: A
Bayesian age-period-cohort modeling study. Cancer Epidemiol. Biomark. Prev. 32, 825–833 (2023).
9. Tanaka, S., Palmer, M. & Katanoda, K. Trends in cervical cancer incidence and mortality of young and middle adults in Japan.
Cancer Sci. 113, 1801–1807 (2022).
10. Wojtyla, C., Janik-Koncewicz, K. & La Vecchia, C. Cervical cancer mortality in young adult European women. Eur. J. Cancer 126,
56–64 (2020).
11. Liu, Q. et al. Development and validation of a seer-based prognostic nomogram for cervical cancer patients below the age of 45
years. Bosn. J. Basic Med. Sci. 21, 620–631 (2021).
12. Kong, Y., Zong, L., Yang, J., Wu, M. & Xiang, Y. Cer vical cancer in women aged 25 years or younger: A retrospective study. Cancer
Manag. Res. 11, 2051–2058 (2019).
13. Pan, S., Jiang, W., Xie, S., Zhu, H. & Zhu, X. Clinicopathological features and survival of adolescent and young adults with cervical
cancer. Cancer Control 28, 1399497690 (2021).
14. Isla-Ortiz, D. et al. Cervical cancer in young women: Do they have a worse prognosis? A retrospective cohort analysis in a
population of Mexico. Oncologist 25, e1363–e1371 (2020).
Fig. 6. ROC curves of the models and FIGO stage for predicting 3-year (a) and 5-year (b) OS in the external
validation cohort.
Scientic Reports | (2024) 14:29808 9
| https://doi.org/10.1038/s41598-024-81644-z
www.nature.com/scientificreports/
Content courtesy of Springer Nature, terms of use apply. Rights reserved
15. Lau, H. Y. et al. Aggressive characteristics of cervical cancer in young women in Taiwan. Int. J. Gynaecol. Obstet. 107, 220–223
(2009).
16. Rutledge, F. N. et al. Youth as a prognostic factor in carcinoma of the cervix: A matched analysis. Gynecol. Oncol. 44, 123–130
(1992).
17. Delaloye, J. F., Pampallona, S., Coucke, P. A. & De Grandi, P. Younger age as a bad prognostic factor in patients with carcinoma of
the cervix. Eur. J. Obstet. Gynecol. Reprod. Biol. 64, 201–205 (1996).
18. Chetty, R. et al. e association between income and life expectancy in the United States, 2001–2014. JAMA 315, 1750–1766
(2016).
19. Harrell, F. E. Jr. Regression Modeling Strategies: With Applications to Linear Models, Logistic and Ordinal Regression, and Survival
Analysis | Springerlink (2001).
20. Li, Z., Lin, Y., Cheng, B., Zhang, Q. & Cai, Y. Prognostic model for predicting overall and cancer-specic survival among patients
with cervical squamous cell carcinoma: A seer based study. Front. Oncol. 11, 651975 (2021).
21. Feng, Y. et al. Nomograms predicting the overall survival and cancer-specic survival of patients with stage IIIC1 cervical cancer.
BMC Cancer 21, 450 (2021).
22. Gauthier, J., Wu, Q. V. & Gooley, T. A. Cubic splines to Model relationships between continuous variables and outcomes: A Guide
for clinicians. Bone Marrow Transpl. 55, 675–680 (2020).
23. Heinzl, H. & Kaider, A. Gaining more exibility in cox proportional hazards regression models with cubic spline functions.
Comput. Methods Progr. Biomed. 54, 201–208 (1997).
24. Pelkofski, E. et al. Cervical cancer in women aged 35 years and younger. Clin. er. 38, 459–466 (2016).
25. Ruiz, R. et al. Clinical-pathological features and survival in young women with cervical cancer: A retrospective analysis from the
Instituto Nacional De Enfermedades Neoplasicas. Rev. Peru Med. Exp. Salud Publica. 34, 218–227 (2017).
26. Higgins, G. D. et al. Increased age and mortality associated with cervical carcinomas negative for human papillomavirus rna.
Lancet 338, 910–913 (1991).
27. Arezzo, F. et al. HPV-negative cervical cancer: A narrative review. Diagnostics 11, 952 (2021).
28. Macios, A. & Nowakowski, A. False negative results in cervical cancer screening-risks, reasons and implications for clinical
practice and public health. Diagnostics 12, 1508 (2022).
29. Itarat, Y. et al. Sexual behavior and infection with cervical human papillomavirus types 16 and 18. Int. J. Womens Health 11,
489–494 (2019).
30. Zhong, X. et al. Prognostic nomogram for rectal Cancer patients with Tumor deposits. Front. Oncol. 12, 808557 (2022).
31. Deng, G. C. et al. Nomogram to predict survival of patients with Advanced and metastatic pancreatic Cancer. BMC Cancer 21,
1227 (2021).
32. Xu, Z. P., Gong, J. H., Gong, J. P. & Li, J. H. A nomogram predicting the outcome of gallbladder cancer patients with dierent target
organ metastases. Eur. Rev. Med. Pharmacol. Sci. 27, 10016–10030 (2023).
33. Pachigolla, S. L. et al. A nomogram predicting early cervical cancer distant recurrence. Int. J. Radiat. Oncol. Biol. Phys. 111, e619–
e620 (2021).
34. Kim, D. Y. et al. Preoperative nomogram for the identication of lymph node metastasis in early cervical cancer. Br. J. Cancer 110,
34–41 (2014).
35. Xie, G. et al. Calculating the overall survival probability in patients with cervical cancer: A nomogram and decision curve analysis-
based study. BMC Cancer 20, 833 (2020).
36. Jiang, K., Ai, Y., Li, Y. & Jia, L. Nomogram models for the prognosis of cervical cancer: A seer-based study. Front. Oncol. 12, 961678
(2022).
37. Zheng, R. R. et al. Nomogram predicting overall survival in operable cervical cancer patients. Int. J. Gynecol. Cancer 27, 987–993
(2017).
38. Chen, W. et al. Nomogram for prognosis of elderly patients with cervical cancer who receive combined radiotherapy. Sci. Rep. 13,
13299 (2023).
39. Xiang, Y., Wang, F., Zhao, M. & Ling, X. Nomograms prediction for young and middle-aged patients (aged under 50) with
metastatic cervical cancer: Based on seer database. Asian J. Surg. 46, 3252–3254 (2023).
40. Zhong, C. et al. A nomogram and risk classication system forecasting the cancer-specic survival of lymph- node- positive rectal
cancer patient aer radical proctectomy. Front. Oncol. 13, 1120960 (2023).
41. Grigsby, P. W. et al. Figo 2018 staging criteria for cervical cancer: impact on stage migration and survival. Gynecol. Oncol. 157,
639–643 (2020).
42. Soares, L. C., de Souza, R. J. & Oliveira, M. Reviewing Figo 2018 cervical cancer staging. Acta Obstet. Gynecol. Scand. 102, 1757–
1758 (2023).
Author contributions
D.Z. contributed to the conception of the study and review of the nal manuscript; Q.Q. contributed signi-
cantly to data collection; Y.G. and W.Z. performed the data analyses and wrote the manuscript; F.G. and W.T.
performed the analysis with constructive discussions. All authors approved the nal manuscript.
Declarations
Competing interests
e authors declare no competing interests.
Additional information
Supplementary Information e online version contains supplementary material available at h t t p s : / / d o i . o r g / 1
0 . 1 0 3 8 / s 4 1 5 9 8 - 0 2 4 - 8 1 6 4 4 - z .
Correspondence and requests for materials should be addressed to D.Z.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Scientic Reports | (2024) 14:29808 10
| https://doi.org/10.1038/s41598-024-81644-z
www.nature.com/scientificreports/
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Open Access is article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives
4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in
any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide
a link to the Creative Commons licence, and indicate if you modied the licensed material. You do not have
permission under this licence to share adapted material derived from this article or parts of it. e images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated
otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence
and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to
obtain permission directly from the copyright holder. To view a copy of this licence, visit h t t p : / / c r e a t i v e c o m m o
n s . o r g / l i c e n s e s / b y - n c - n d / 4 . 0 / .
© e Author(s) 2024
Scientic Reports | (2024) 14:29808 11
| https://doi.org/10.1038/s41598-024-81644-z
www.nature.com/scientificreports/
Content courtesy of Springer Nature, terms of use apply. Rights reserved
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com
