In our study, based on univariate cox proportional hazards regression, we found that age, gender, anatomical site, T stage, N stage, M stage, AJCC stage, SEER stage, Breslow depth, ulceration and treatment were associated with OS and CSS in NM patients. However, in multivariable cox proportional hazards regression, T and N stages were not strongly associated with prognostic factors for OS and CSS rates in NM patients. At the same time, we constructed a nomogram to quantitatively predict individual 3- and 5-year OS and CSS rates through patient-related and tumor-related factors. Our nomograms can be used to estimate patients’ outcomes and make individualized decisions about monitoring and treatment.
Most studies have confirmed that advanced age is a relevant risk factor for the prognosis of NM, which is consistent with the results obtained in our study. One study analyzed the peak age of nodular melanoma at 65 to 69 years of age. However, the age cutoffs varied across studies. Therefore, in this study, the age was divided into < 70 years, 70–84 years and > 84 years by X-Tile software as cut-off points, which could better distinguish the survival rate of some variables. A study of 3076 patients7 found that with increasing age, Breslow depth, the presence of ulceration, and the proportion of males also increased, suggesting that age is associated with other risk factors. Complications in the elderly population may also be an important factor affecting the prognosis of melanoma, but other complications of patients were not recorded in the SEER database, so it was not possible to analyze and study this, which is also one of the limitations of this study. At the same time, we found that gender was also a relevant prognostic factor, with most studies showing that being female was a protective factor for melanoma survival, with higher 3- and 5-year survival in women. One study8 found that the 5- and 10-year overall survival rates for men were 69% and 61%, respectively, compared with 82% and 75% for women. This is similar to our findings. Race, which has been shown in other studies to be associated with melanoma prognosis9. But this was not reflected in our study. We believe that this is related to the fact that the majority of patients included in the database are Caucasian, and we also consider that different subtypes of melanoma may behave differently for an independent factor of race. Other studies have analyzed differences in survival between different ethnic groups and have been correlated with melanocyte distribution, sunlight exposure time, and genetic mutations in different ethnic groups. Further studies are needed to explore the relationship between survival and ethnicity in different subtypes of melanoma10.
Location is also a fairly important prognostic factor in some studies, with extremities having better survival than trunk locations11. However, other studies have shown that the survival rate of the proximal site is worse than that of the distal site12. Most studies concluded that melanoma located in the head and neck had a worse prognosis, with a 1.84-fold higher risk of death in patients with CMM of the head and neck compared with patients of the extremities in multivariate analysis13. In our study, NM with primary sites of the trunk and extremities had a higher survival rate, and NM in the head and neck site had a lower survival rate, similar to the findings of some of these studies. Studies have confirmed that the thickness of the tumor, that is, the Breslow depth, is directly proportional to the mortality of melanoma, and the deeper the Breslow depth, the higher the positive rate of sentinel lymph node biopsy, that is, the easier it is to metastasize14,15. This is the same result as our study. Ulceration are an important factor in rapid tumor growth secondary to ischemia, so tumors with ulceration are thought to be more aggressive16. Most studies have shown that ulceration is a significant predictor of reduced overall survival17 and a significant predictor of recurrence18 in multivariate analysis. This was also verified in our study. The T, N, M stage, AJCC stage and SEER stage mentioned in our study are also important prognostic factors after univariate analysis, but T and N stage have little significance in multivariate analysis. The results were slightly different, and further research is needed. This is similar to the results of previous studies. Surgery is the main treatment for melanoma. Surgical methods include local excision, local wide excision, lymph node dissection, or sentinel lymph node biopsy19. At present, a variety of new treatment options have emerged, and adjuvant therapy is recommended for some patients with stage II and patients with stage III and IV. However, currently used chemotherapy drugs are of limited value for most patients with stage IV melanoma. Adjuvant therapies include: immunotherapy (melanoma vaccine, interferon, interleukin-2), chemotherapy, radiotherapy, and biological chemotherapy. Since no other new treatment options were included in the SEER database, we used surgical treatment as the analysis indicator. Other studies have also confirmed that surgical treatment is a good prognostic factor, which is the same as our findings.
In this study, we used a huge database, adding more possible prognostic related factors such as: age, sex, race, anatomical site, TNM stage, SEER stage, AJCC stage, depth, ulcer and treatment, these factors are readily available in clinical work and better represent the clinical and pathological features of NM. Our nomogram shows good discriminative power in predicting prognosis, and is more comprehensive than previously constructed nomograms. In the present study, both the internal and external C-index were above 0.78, showing a pleasing discriminative ability to provide patients with prognostic information in a personalized manner. Likewise, AUCs also implies good discriminative ability. The calibration curve shows that the predicted values of the nomogram have high agreement. In addition, DCA was performed to provide the clinical net benefit of the predictive model. In the present study, all results indicated that the DCA curves of the 3- and 5-year OS rates of the new model yielded a significant net clinical benefit.
Our study has some limitations. First, the nomogram data comes from the SEER database, which contains only about 30% of the U.S. population, which is mostly Caucasian. Therefore, ethnicity and population are more limited, and although the data volume is large, the population category is relatively single. We need to augment data from other populations to more fully characterize the disease in NM. These should be factored into future forecasting models. Second, we did not include other prognostic factors such as marital status, economic status, education, complications, novel treatment modalities, mitotic index, presence and number of tumor-infiltrating lymphocytes (TILs), Capillary invasion, presence of microscopic satellites, and presence of perineural invasion. vi If this information is incorporated into the model, it will make nomogram predictions more accurate and personalized in the future. Finally, patients were divided into two groups, 70% of which were used for construction and the remaining 30% for validation of nomograms. The C-index, AUC, calibration curve and DCA performed well, but further studies are needed to externally validate the proposed nomogram.