Introduction

Lung cancer is the leading cause of cancer-related death and the second most commonly diagnosed malignancy worldwide [1]. Currently, adenocarcinoma (ADC) is the most common histologic type. In 2011, International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society (IASLC/ATS/ERS) further classified lung adenocarcinoma into five pathological subtypes: lepidic, acinar, papillary, micropapillary and solid patterns [2]. Micropapillary and solid components were considered to be high risk and closely associated with poorer prognosis [3, 4].

Lobectomy with mediastinal lymphadenectomy has been recommended as the standard surgical treatment of early-stage non-small cell lung cancer (NSCLC) for a long time. While recent studies have confirmed segmentectomy could be another option for tumors of 2 cm or smaller [5, 6]. However, Nitadori and colleagues proved that for tumors with MIP component ≥ 5%, patients treated with lobectomy had a reduced risk of recurrence compared with those treated with limited resection [7]. A similar conclusion had been reached by Su and colleagues. For lung ADCs presenting as pure solid nodules, segmentectomy was significantly associated with worse recurrence-free survival (RFS) and overall survival (OS) in patients with MIP component > 5% than lobectomy [4). Finally, we evaluated the discrimination of the model by ROC curve and certain discrimination was found with an area under curve (AUC) of 0.893 (95%CI 0.853–0.932) (Fig. 5). It means the model can discern a patient with an event from a patient without an event 89.3% of the time.

Fig. 3
figure 3

The nomogram to predict the probability of S/MP5 + in small-sized lung adenocarcinoma

Full size image
Fig. 4
figure 4

The calibration plot of the nomogram

Full size image
Fig. 5
figure 5

The ROC curve of the model

Full size image

For every patient, the value of each variable was given a score on the point scale axis. The total score could be calculated by adding every single score together. Locate it on the total points axis and then draw a line straight down to get the exact probability of S/MP5+.

The X-axis and Y-axis respectively represented the predicted and actual probability of S/MP5+. A perfect prediction model would correspond to the black dotted line. The red and blue full line respectively represented the apparent derived from the entire cohort and bias-corrected curve by bootstrap**.

Discussion

In recent years, segmentectomy is gradually accepted as an alternative surgical procedure for lobectomy of small-sized NSCLC. However, previous studies have confirmed for adenocarcinoma ≤ 2 cm with 5% or more high-grades subtypes components, segmentectomy is associated with higher frequency of recurrence compared with lobectomy. Thus, lobectomy still should be more considered for this part of patients. Therefore, it’s of great value to predict whether high-grade subtypes components higher than 5% in small-sized adenocarcinoma preoperatively. In this retrospective study, we built a nomogram to try to solve this problem based on CT diameter, CTR and SII. The nomogram demonstrated relatively good performance through calibration plot and ROC curve. For patients with a solitary nodule smaller than 2 cm suspected of lung adenocarcinoma, the model may be useful for selecting the optimal surgical procedure from the perspective of pathological subtypes components. However, the limitations of our model should also be addressed. Nomogram itself has the shortcoming of overfitting. Thus, the discrimination can vary greatly if applied to different cohorts, despite excellent discrimination was observed in our derivation cohort. Therefore, the generalizability of our model still need external validation using data from other centers.

Compared with S/MP5- patients, S/MP5+ patients had a higher frequency of pleural invasion, lymphovascular invasion, lymph node metastasis and STAS, which was consistent with the results -of previous studies [9, 17, 18]. Therefore, relatively poorer prognosis of S/MP5+ patients was foreseeable due to differences in these high-risk factors even though long-term follow-up has not been completed yet.

Most of the patients (72.1%) enrolled in the present study underwent lobectomy. However, further analysis of patients with different high-grade subtypes components revealed that a significantly higher proportion of S/MP5+ underwent lobectomy compared with S/MP5- patients. We supposed it was mainly related to the larger tumor sizes and the higher CTR. Lobectomy might be more likely selected considering the higher rate of invasive adenocarcinoma and poorer prognosis [19, 20]. However, from the perspective of postoperative pathology, our selection of surgical approaches remained conservative. 176 (67.2%) S/MP5− patients underwent lobectomy while segmentectomy might be the optimal choice for these patients. On the other hand, 9 patients underwent segmentectomy but confirmed as S/MP5+. Lobectomy might be more beneficial for improving the prognosis of these patients.

The most important predictor in our model was SII. Nevertheless, it had some inevitable limitations due to its role as an inflammatory index. It is greatly affected by infections, drugs, other malignancies, hematological disease or even exercise and weather [21, 22]. Thus, combining with other possible indicators may further improve predictive accuracy and practicality. Frozen section (FS) is another commonly used means to identify pathological subtypes during the operation. However, its sensitivity of identifying minor high-grade subtypes components was low according to previous studies [23, 24]. Moreover, obtaining tissue for FS is difficult in many cases limited by the location of the lesion. Imaging is another promising approach. We have found CTR which based on CT an independent predictor of S/MP5+ in the present study. 18F-fluorodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT) is a commonly used tool in the management of cancer patients. A study conducted by Chang and colleagues revealed that for lung adenocarcinoma ≤ 3 cm, maximum standard uptake value (SUVmax), metabolic tumor volume (MTV) and total lesion glycolysis (TLG) were all significantly higher in patients with minor high-grade subtypes components than negative patients [25]. However, due to the low incidence of lymph node or distant metastasis and the expensive fees, PET/CT is rarely performed in patients with tumors less than 2 cm in our institution. Therefore, other simple imaging features to predict subtypes components still requires further investigation.

In 2020, a new grading system for invasive pulmonary adenocarcinoma was proposed by the International Association for the Study of Lung Cancer Pathology Committee [26]. In this new grading system, lung adenocarcinomas were classified into 3 grades based on the predominant and high-grade histologic pattern with a cutoff of 20%. However, we didn’t apply it as our theoretical basis mainly because of its different inclusion criteria. Stage I cases were enrolled in that study. Thus, some tumors between 2 and 4 cm might be included while cases with lymph node metastasis were excluded. Besides, the study did not provide surgical approaches for patients with different grades. Therefore, we still based our research on high-grade patterns with a cutoff of 5%.

Inflammation serves an important role in tumor progression, invasion and metastasis [27]. Neutrophils promote tumor progression by releasing angiogenic factors, including vascular endothelial growth factor (VEGF), matrix metalloproteinase 9 (MMP-9) and prokineticin-2 (PROK2) [28]. Platelets have been proven to protect tumor cells from shear forces and assault of NK cells and stimulate tumor cell proliferation to form micro metastasis foci. Meanwhile, platelet-derived growth factors help to open the capillary endothelium to accelerate tumor cell extravasation [29]. Thus, neutrophils and platelets are considered primarily pro-tumor. On the contrary, lymphocytes are mainly responsible for combating external infection, clearing variant cells in the body, and exerting inhibitory effects on tumor generation and progression. Hence, either an increase in the number of platelets or neutrophils or a decrease in the number of lymphocytes leads to an elevation of SII, indicating higher invasiveness of the tumor, consistent with the general cognition of each pathological subtype of lung adenocarcinoma. However, the specific mechanisms of why solid or micropapillary components cause such changes are not very clear and require further exploration.

In addition to the shortcomings of nomogram itself mentioned before, several limitations should also be pointed out in our study. First of all, the retrospective nature and limited samples of our study therefore may introduce some inevitable selection bias. Besides, we just analyzed total lymphocyte count to calculate SII rather than lymphocyte subsets. It has been proved that CD3 + T lymphocytes, CD4 + T lymphocytes and the rate of CD4/CD8 are closely correlated with the prognosis of lung cancer [30]. We suspect more satisfactory results may be obtained by applying the data of each subpopulation. In addition, we didn’t perform prognostic follow-up for the included patients due to our small sample size, especially the small number of patients who underwent segmentectomy in the S/MP5+ group. Above all, multicenter researches with larger sample size are needed to further optimize our model and improve its practicality in the future. On the other hand, the specific mechanism by which the increase in SII and the presence of high-grade subtypes components is still unclear. Further researches on lymphocyte subsets or analysis of tumor microenvironment in tumor specimens may be helpful.

Conclusions

High-grade subtypes components are important in determining the surgical approach for small-sized adenocarcinoma. SII, CTR and diameter was three simple predictors of the presence of 5% or more micropapillary or solid patterns. The nomogram constructed based on them had good efficacy to predict it preoperatively.