As a type of non-small cell lung cancer (NSCLC), pulmonary sarcomatoid carcinomas (PSCs) has the characteristics of low incidence, early metastasis and low survival rate[6, 7].Giant cell carcinoma of the lung (GCCL) is one of the pathological subtypes of PSCs. Due to its low incidence, there are few studies on GCCL. Currently, most studies on GCCL are case reports, and population-based studies are lacking. As a prediction model, nomogram has been widely used to predict the prognosis of cancers. Many nomograms are available for NSCLC[15, 18, 22, 23], but none of them focus on GCCL. Many studies focusing on PSCs and suggest that distant metastasis, tumour size, complete resection, TNM staging, tissue subtype were independent prognostic factors[24, 25]. However, there is no research on the prognostic prediction model of GCCL.
In this study, more than 400 cases with GCCL were enrolled, these cases were randomly divided into training cohort and validation cohort which were used to construct and validate the nomogram. Through univariate analysis, the statistically significant variables were selected and included in multivariate analysis, and this nomogram was constructed based on the Cox proportional hazards model. Finally, nine variables, including age, marital status, site, T stage, N stage, M stage, surgery, chemotherapy and radiation therapy were selected as significant prognostic factors for the construction of nomogram. Validation of the model showed its great performance in predicting the prognosis of GCCL. Subsequently, the calibration curve and DCA both confirmed that this nomogram has a good performance in predicting the prognosis of GCCL.
In this study, Table 1 shows the clinicopathological characteristics of GCCL. It can be seen that for patients with GCCL, Patients over 70 years old account for 35.5%, and most of them are male (63.3 %). This is consistent with the conclusions of several studies on PSCs[7, 26]. Among all the cases included in the study, most patients were married patients (55.5%), and white people accounted for the majority (78.4%). GCCL is mostly found in the upper lobe (57.3%), and the right lung is more prevalent than the left (56.2% vs 41.4 %). This is also in line with prior findings on PSCs[6, 7, 26], implying that the clinical features of GCCL are similar to those of PSCs.
Regarding the pathological characteristics of GCCL, Table 1 shows that patients with T4 accounted for 42.4% of GCCL patients, and the majority of patients (61.5%) had lymph node metastases or distant metastasis (41.3%). This is consistent with the conclusions of several studies on PSCs: PSCs is a highly aggressive tumor that is generally large in size and accompanied by metastases when discovered[24, 27-29]. Figure 2 shows that the 1-year and 5-year OS of GCCL are only 0.316 and 0.163. In 2015, Li et al. reported 38 patients with PSCs, 1-year and 5-year OS were 68.4% and 18.4%. In 2018, Mehrad et al. reported 53 cases of surgically treated PSCs, 5-year overall survival was 12.5%. Other studies on PSCs suggest that the 5-year OS is about 10-20%[32, 33]. The prognosis for GCCL is similar with prior PSCs research findings. We see that GCCL is a highly malignant lung cancer with poor prognosis.
According to the nomogram (Figure 3), a total of 9 prognostic factors for predicting GCCL were included. Patients over the age of 70 have a poor prognosis. The prognosis of married patients is better than that of single patients, which is consistent with the findings of several research on breast cancer, prostate cancer and colorectal cancer, which reveal that married cancer patients have a better prognosis [34, 35]. According to some study, the influence of marital status on tumor prognosis is attributable to married patients' better mood and quality of life. According to the nomogram, tumour site is also a prognosis factor of GCCL. Lung NOS has the greatest impact on the prognosis. This may be related to the patients whose tumour location is registered as Lung NOS in the SEER database are the patients with complex tumour conditions such as multiple tumours or located in the main trachea. In addition, we see that prognosis of the lower lobe is worse than that of the upper lobe and middle lobe. There is no relevant research to confirm this conclusion. More research into the relationship between lung tumor site and prognosis is needed in the future to confirm whether pulmonary tumors in the lower lobe have a worse prognosis.
According to the nomogram in this study, T stage, N stage and M stage all affect the patients’ OS, but for T stage and N stage, it can be seen that the difference between T3 and T4 on the prognosis is negligible, and prognosis of T4 is slightly better than T3. The prognosis of cases with lymph node metastasis is worse than that of cases with no lymph node metastasis, but there are small differences in prognosis between N1, N2, and N3, and the prognosis of N2 is slightly better than that of N1, suggesting that the traditional TNM stage may not be applicable for GCCL. More research is needed to improve the staging rules for GCCL.
According to the nomogram, it is obvious that surgery has the greatest impact on the prognosis in the treatment of GCCL. Surgical treatment can significantly improve the prognosis. This is consistent with the current view that surgery is the most effective treatment for PSCs[7, 26]. Furthermore, the nomogram shows that chemotherapy can enhance the patient's OS greatly, however some recent research on PSCs reveal that the therapeutic efficacy of chemotherapy is still debatable [24, 37, 38]. Looking forward to more studies on GCCL in the future to clarify the effect of chemotherapy as a treatment method for GCCL. According to the prediction model constructed in this study, the impact of radiation on GCCL can be demonstrated to be limited. Therefore, for patients with GCCL, surgery is highly recommended. Despite the fact that the current therapeutic impact of chemotherapy on PSCs is controversial, chemotherapy, according to this study, can greatly improve the prognosis of GCCL, making it a recommended treatment option. Radiotherapy is not the first choice of treatment for GCCL because it has little influence on the prognosis.
This research still has certain limitations. First, selection bias is inevitable because this is a retrospective study. Second, because the information originated from the SEER database, certain information that could have influenced the prognosis, such as comorbidities, surgical complications, and chemotherapy regimen was not included. Third, due to the low incidence of GCCL, sufficient external data for validation in this study was not available. More prospective multicentre trials on GCCL are likely to be undertaken in the future to improve the diagnosis and treatment of this disease.