Keywords
metastases/metastasectomy - lung cancer - diagnosis - histology
Introduction
The lung is the second most common site of colorectal cancer (CRC) metastases, and colorectal metastases are the most common indication for pulmonary metastasectomy (PM).[1] Many retrospective studies demonstrated that PM can lead to improved disease control and prolonged survival in carefully selected patients. The reported post-PM 5-year survival rate is approximately 50%, which is significantly better than that predicted for metastatic CRC and indicates that resection provides a survival benefit.[2] Surgery is the standard treatment for patients with isolated lung metastases,[3] but local recurrence within the surgical margin following a limited surgical approach is a known problem, with relatively high incidence.[4]
[5] One of the most important factors determining the occurrence of local recurrence in the surgical margin in patients with pulmonary CRC metastases is the number of aerogenic spreads with floating cancer cell clusters,[4]
[5] which was later named “spread through air spaces (STAS)” in primary lung cancer.[6] STAS has been recognized as a new pattern of invasion in lung adenocarcinoma and its role in worsening prognosis and increasing the likelihood of recurrence was postulated.[7] In published studies, the classification and definition of STAS varies in terms of distance from the main tumor and morphologic features.[8] The aim of this study is to further investigate the prevalence and prognostic significance of STAS in lung metastases from CRC.
Patients and Methods
All patients who underwent anatomical (lobectomy/segmentectomy) and nonanatomical (wedge) pulmonary resection with curative intent for solitary pulmonary metastases from CRC (SPM-CRC) at our single institution between 2008 and 2015 were retrospectively analyzed. Selection criteria for surgery were as follows: stable primary tumor site, intent of complete (R0) resection, and cardiorespiratory fitness. Stable tumor site meant either the absence of a primary tumor, for example, after resection or definitive radiochemotherapy applied, or the absence of primary tumor progression according to standard oncological follow-up criteria at the time the indication for PM was made. All metastatic lesions were evaluated by a senior pathologist to ensure complete (R0) resection.
Preoperative patient characteristics are outlined in [Table 1].
Table 1
Patient characteristics after PM for solitary colorectal metastasis
|
Characteristics
|
Number of patients
|
%
|
|
Total
|
49
|
|
|
Gender
|
|
Male
|
30
|
61.2
|
|
Female
|
19
|
38.8
|
|
Primary tumor site
|
|
Colon
|
14
|
28.6
|
|
Rectum
|
35
|
71.4
|
|
Synchronous metastases
|
|
No
|
45
|
91.8
|
|
Yes
|
4
|
8.2
|
|
Extrapulmonary metastases
|
|
No
|
34
|
69.4
|
|
Yes
|
15
|
30.6
|
|
Location of pulmonary metastasis in CT
|
|
Central
|
5
|
10.2
|
|
Peripheral
|
42
|
85.7
|
|
Missing
|
2
|
4.1
|
|
Presence of STAS
|
|
No
|
36
|
73.5
|
|
Yes
|
13
|
26.5
|
|
Surgical technique
|
|
Open
|
41
|
83.7
|
|
VATS
|
8
|
16.3
|
|
Type of pulmonary resection
|
|
Anatomical resection
|
24
|
49.0
|
|
Wedge resection
|
25
|
51.0
|
|
Adjuvant therapy for primary tumor
|
|
No
|
11
|
22.4
|
|
Yes
|
38
|
77.6
|
|
Induction therapy prior to PM
|
|
No
|
43
|
87.8
|
|
Yes
|
6
|
12.2
|
Abbreviations: CT, computed tomography; PM, pulmonary metastasectomy; STAS, spread through air spaces; VATS, video-assisted thoracic surgery.
Pathological Analysis
The histopathology slides of resected SPM-CRC were reassessed by an experienced pathologist blinded to the patients' clinical data [Fig. 1]. The different diagnostic criteria for STAS in the various studies may have contributed to numerous discrepancies in the pathological identification of STAS. Onozato et al[9]
[10] used the term “tumor island” to refer to isolated, large collections of tumor cells within alveolar spaces without well-defined configuration.[8] Warth et al[11] classified tumor cells as STAS when there is no direct connection of the cells to the main tumor mass. Warth et al[11] have also categorized STAS according to their distance from the primary tumor as limited when the solid cell nests are no more than 3 alveoli away from the primary tumor and as extensive when they are more than 3 alveoli away.
Fig. 1(A–C) Spread through air spaces (STAS) cluster (as indicated by the arrowheads) detected in the air spaces of the lung parenchyma beyond the margin of the main tumor. (C) Noted a reactive response in the intra-alveolar space with an isolated tumor island (presence of lymphocytes and fibroblasts).
In our study, STAS was defined morphologically as the presence of tumor cells without direct connection to the main tumor mass. To reduce the risk of including artificially dissected cells during metastasis resection or grossing, an accumulation of more than five tumor cells in the alveolar space was considered STAS.[5]
[11] However, we identify isolated, large cluster of tumor cells “Tumor Island” as a subtype of STAS. In regards to the distance parameter, STAS was located in intra-alveolar spaces within 0.5 mm of the main tumor margin. If the alveoli are collapsed (atelectatic), for example, we considered STAS from the third alveolar space from the main tumor as a method to measure the distance. Slides were digitized and reevaluated using a virtual microscope.
Analysis of Computed Tomography Images and Locoregional Recurrence
The central versus peripheral location of pulmonary metastases was assessed on preoperative chest computed tomography (CT) and defined according to the American College of Chest Physicians (ACCP), National Comprehensive Cancer Network (NCCN), and European Society of Thoracic Surgery (ESTS) guidelines. The ACCP defines central lesions as those located in the inner third of the hemithorax, whereas the NCCN and ESTS define central lesions as those located in the inner two-thirds of the hemithorax.[12] In this study, we decided to apply the NCCN and ESTS definition of central lesion. Postoperative chest CT images were reviewed for locoregional recurrence. Recurrence was classified either as locoregional, or as tumor evidence within the same segment/lobe or ipsilateral side of the original metastatic lesion. All lesions radiologically suspected for recurrence were qualified for resection. [Fig. 2] demonstrates the thirds concept of hemithorax.
Fig. 2 (A) The (I) lines represent the thirds of the hemithorax that are straight lines on a coronal computed tomography (CT) image of the thorax parallel to the chest wall. (B) The (II) lines also represent the thirds of the hemithorax seen on an axial CT image of the thorax as eccentric circles emanating from the hilum.
Statistical Analysis
Categorical variables including gender, location of STAS, and other specified factors, were assessed using the chi-square test to analyze the relationship between STAS and locoregional recurrence of metastases. The significance of differences between the two groups (locoregional recurrence and number of the tumor-surrounding alveoli affected by STAS) was assessed using the Mann–Whitney test.
The effect of age, disease-free interval (DFI), and size of pulmonary metastasis on the risk of STAS was assessed using three logistic regression models in which the dependent variable was the presence of STAS and the independent variables were, in turn, age, DFI, and size of pulmonary metastasis. The results were grouped with the mean and standard deviation of each factor in relation to the presence of STAS, the odds ratio of the presence of STAS with a 1-year change, a 1-month change in DFI, and a 1-cm change in metastasis size, and the p-value of the test probability to assess the significance of the effect of each factor on the presence of STAS. The odds ratio values of STAS for the comparison groups of a given factor (together with the 95% confidence interval) were calculated. For both univariable and multivariable analysis, a probability value (p) of less than 0.05 was considered statistically significant. STATISTICA 10.0 (StatSoft, Tulsa, Oklahoma, United States) software was used for statistical analysis.
Results
Forty-nine patients (30 men and 19 women) were included in the study. The presence of STAS was confirmed in 26.5% (n = 13) patients. STAS was detected in 14.3% (2 of 14) patients with colon carcinoma compared with 31.4% (11 of 35) patients with rectal carcinoma. Although there was a clear difference between these site-dependent frequencies, it was not statistically significant for STAS (p = 0.2195). The majority of patients (83.7%, n = 41) underwent open surgery, 16.3% (n = 8) of patients underwent video-assisted thoracic surgical resections, with no statistical significance for STAS (p = 0.325). Twenty-five patients (49%) underwent wedge resection, while anatomical resections were performed in 51% (n = 24) of patients including segmentectomy (n = 13), lobectomy (n = 10), and bilobectomy (n = 1). There was no correlation between STAS and extent of resection (anatomical vs. wedge, p = 0.376). Resections of metastatic lesions were performed using staplers and cautery. Laser resections were not performed in the presented group. Systematic lymph node dissection was not performed routinely. The association between STAS and clinical characteristics of patients after resection of pulmonary metastases was evaluated and summarized in [Table 2].
Table 2
Relationship between STAS and clinical characteristics of patients after PM
|
Variables
|
STAS-positive
number
|
%
|
OR (95% CI)
|
p-Value
|
|
Gender
|
2.67 (0.63–11.35)
|
0.175
|
|
Male
|
10 (n = 30)
|
33.3
|
|
|
|
Female
|
3 (n = 19)
|
15.8
|
|
|
|
Primary tumor site
|
2.75 (0.52–14.44)
|
0.219
|
|
Colon
|
2 (n = 14)
|
14.3
|
|
|
|
Rectum
|
11 (n = 35)
|
31.4
|
|
|
|
Synchronous metastasis
|
0.92 (0.09–9.69)
|
0.942
|
|
No
|
12 (n = 45)
|
26.7
|
|
|
|
Yes
|
1 (n = 4)
|
25.0
|
|
|
|
Extrapulmonary metastasis
|
0.60 (0.14–2.60)
|
0.491
|
|
No
|
10 (n = 34)
|
29.4
|
|
|
|
Yes
|
3 (n = 15)
|
20.0
|
|
|
|
Location of pulmonary metastasis in CT
|
52.07 (2.59–1046.1)
|
0.0001
|
|
Central
|
5 (n = 5)
|
100
|
|
|
|
Peripheral
|
7 (n = 42)
|
17
|
|
|
|
Surgery technique
|
2.90 (0.32–26.16)
|
0.325
|
|
Open
|
12 (n = 41)
|
29.3
|
|
|
|
VATS
|
1 (n = 8)
|
12.5
|
|
|
|
Type of PM
|
1.79 (0.49–6.53)
|
0.376
|
|
Anatomical resection
|
5 (n = 24)
|
20.8
|
|
|
|
Wedge resection
|
8 (n = 25)
|
32.0
|
|
|
|
Adjuvant therapy for primary tumor
|
0.95 (0.21–4.31)
|
0.949
|
|
No
|
3 (n = 11)
|
27.3
|
|
|
|
Yes
|
10 (n = 38)
|
26.3
|
|
|
|
Induction therapy
|
1.45 (0.23–9.07)
|
0.687
|
|
No
|
11 (n = 43)
|
25.6
|
|
|
|
Yes
|
2 (n = 6)
|
33.3
|
|
|
Abbreviations: CI, confidence interval; CT, computed tomography; OR, odds ratio; PM, pulmonary metastasectomy; STAS, spread through air spaces; VATS, video-assisted thoracic surgery.
Clinical-Pathological Findings and Locoregional Recurrence
The median length of follow-up was 95 months (range 50–163 months). The median size of the resected lesion was 1.4 cm (range 0.6–4.5 cm). The median number of the STAS-involved alveoli was 4 (range 1–9). There was a statistically significant association between the number of tumor-surrounding alveoli invaded by STAS and locoregional recurrence of metastases (p = 0.0001 [Table 4]). [Table 3] demonstrates associations between STAS and other factors.
Table 3
Correlation between STAS and age, size of metastases, and DFI
|
Variable
|
Presence of STAS
|
OR (95% CI)
|
p-Value
|
|
No
|
Yes
|
|
Mean
|
SD
|
Mean
|
SD
|
|
Age (y)
|
65.3
|
11.7
|
60.2
|
10.0
|
0.962 (0.910–1.017)
|
0.1742
|
|
Size of pulmonary
metastasis (cm)
|
1.76
|
1.01
|
1.72
|
1.16
|
0.964 (0.517–1.799)
|
0.9089
|
|
DFI (mo)
|
28.8
|
20.9
|
37.7
|
34.0
|
1.014 (0.989–1.039)
|
0.2755
|
Abbreviations: CI, confidence interval; DFI, disease-free interval; OR, odds ratio; SD, standard deviation; STAS, spread through air spaces.
Table 4
Correlation between number of STAS clusters and locoregional recurrence of metastasis
|
Locoregional recurrence of metastases
|
|
No (n = 45)
|
Yes (n = 4)
|
|
Mean
|
Median
|
SD
|
Mean
|
Median
|
SD
|
p-Value
|
|
No. of STAS clusters
|
0.23
|
0
|
0.84
|
4.00
|
3.5
|
1.41
|
0.0001
|
Abbreviations: SD, standard deviation; STAS, spread through air spaces.
Ninety-six percent (n = 47) of preoperative CT images were available for analysis of the location of metastases (central vs. peripheral). There were 10.2% (n = 5) centrally versus 85.7% (n = 42) peripherally located lesions ([Table 1]). We found that all 5 patients with central metastases were STAS positive compared with 17% (n = 7 of 42) peripheral metastases (p = 0.0001, [Table 2]).
All STAS-negative patients remained recurrence-free. In 30.7% of STAS-positive patients (n = 4 of 13) there was radiological suspicion of locoregional recurrence. All 4 patients with a suspected lesion underwent pulmonary resection, which confirmed the locoregional recurrence. The presence of STAS was therefore significantly associated with locoregional recurrence (p = 0.003). Five-year overall survival rates in 13 STAS-positive patients were not significantly different compared with 36 STAS-negative patients (p = 0.6651, [Fig. 3]).
Fig. 3 Five-year survival rate in spread through air spaces (STAS)-positive versus STAS-negative patients with solitary colorectal lung metastasis.
Discussion
Previous studies reported the incidence of pulmonary metastases of 5.6% in rectal carcinoma and 3.7% in colon carcinoma, almost 5% of CRC patients had synchronous lung metastases.[13]
[14] In other reviews, the incidence of CRC lung metastases ranged from 5 to 15%, including metachronous metastases.[13]
[15] Although controversial according to many authors, so-called curative-intent surgery remains the first-line treatment worldwide for CRC lung metastases.[13]
[16] Several retrospective studies have concluded that PM provides long-term survival benefits, but this has not been supported by the results of randomized trials. A challenge to the curative intent concept of pulmonary metastatic CRC is the recurrence of metastases; numerous retrospective studies have investigated this topic.
Previous studies have shown recurrence of CRC lung metastases in 50% of patients after PM; multiple recurrence prognosticators have been identified: DFI, carcinoembryonic antigen, number of lesions, and thoracic lymph node involvement.[17] STAS is being investigated as one of these prognostic factors.
Some studies described STAS as a mode of neoplastic spread that pulmonary metastases may adopt.[4]
[5] In 2015, STAS was defined as one or more pathological micropapillary clusters, solid nests, or single tumor cells extending beyond the tumor edge into the air spaces of the surrounding lung parenchyma and detaching from the main tumor.[15] The latest 2021 World Health Organization Classification of Thoracic Tumours recognizes STAS as a histological feature of prognostic significance in adenocarcinoma of the lung.[18] In our study of 49 patients with solitary CRC lung metastases, we identified STAS in 26.5% (n = 13) of patients. There was a close association between STAS and the central location of lung metastases on CT images (p = 0.0001). In our study, we aimed to distinguish STAS from artifacts occurred during surgical resection. Several studies reported relapse within the resection margin after pulmonary wedge resection for pulmonary CRC metastases.[4]
[5] We hypothesize that due to the distant presence of STAS beyond the margin of pulmonary metastatic lesion, local recurrence may be more extensive than recurrence only within the surgical resection margin. Nevertheless, all 5 patients with centrally located metastases underwent anatomical resection. Of the 42 patients with peripheral metastases, 19 underwent anatomical resection and 23 underwent wedge resection. Therefore, it is very unlikely that the central lesions had insufficient resection margin.
In our study, STAS was strongly associated with locoregional recurrence and the number of STAS-positive alveoli was a significant factor in determining local recurrence (p = 0.0001). Histological examination showed no association between STAS and the size of lung metastases (p = 0.908).
As most lung tissue consists of air spaces, the likelihood of tissue fragments being displaced and trapped in these spaces during surgery is much higher. The spread of benign or malignant tissue fragments in airspaces may be caused by mechanical stresses at the tissue level as an artifactual effect. This artifact occurs in tumors with poor cohesiveness and fewer intercellular adhesions, which is responsible for the worsened prognosis.[19] On the other hand, we suggest that the tumor microenvironment (TME) may be responsible for the origin of STAS. TME is often associated with tumorigenesis, as it indicates the involvement of tumor cells in surrounding stromal cells, which affects tumor progression. Interestingly, nonmalignant cells in the TME play an important role in all phases of carcinogenesis.[8]
[20]
Importantly for the surgeons, this study did not demonstrate statistical significance between anatomical versus wedge resection in terms of the presence of STAS, nor differences in post-PM overall survival between patients with and without STAS. We believe that the decision between anatomical and wedge resection should not be made solely on the basis on the STAS presence to avoid local recurrence, as metastatic disease may also relapse despite anatomical resection. The incidence of intrathoracic lymph node involvement at the time of PM is higher in CRC compared with other epithelial pathologies, ranging from 12 to 44%. Intrathoracic lymph node involvement has been found to be a negative survival prognosticator for pulmonary metastatic CRC.[21]
[22] However, mediastinal lymph node dissection is not routinely performed during PM.[13]
The limitation of our study is the small size of the presented cohort. However, the broad definition and different diagnostic criteria for STAS in previous studies may lead to different analyses, for example, in our study, we identified the isolated, tumor island as a subtype of STAS. Another limitation is the single-institution retrospective design, for this reason, our analysis should rather be considered as a hypothesis-generating case series study.
Conclusion
Through this study, we were able to investigate the presence of STAS in solitary CRC lung metastases and its prognostic significance by associating it with other clinicopathological factors. Our most important finding is the significant association between centrally located CRC metastases and the presence of STAS.
