Fax: (011) 81-3-3248-2463
Cytopathologic factors can predict invasion in small-sized peripheral lung adenocarcinoma with a bronchioloalveolar carcinoma component
Version of Record online: 9 NOV 2006
Copyright © 2006 American Cancer Society
Volume 108, Issue 6, pages 488–493, 25 December 2006
How to Cite
Maezawa, N., Tsuta, K., Shibuki, Y., Yamazaki, S., Maeshima, A. M., Watanabe, S.-i. and Matsuno, Y. (2006), Cytopathologic factors can predict invasion in small-sized peripheral lung adenocarcinoma with a bronchioloalveolar carcinoma component. Cancer, 108: 488–493. doi: 10.1002/cncr.22287
- Issue online: 11 DEC 2006
- Version of Record online: 9 NOV 2006
- Manuscript Revised: 15 AUG 2006
- Manuscript Accepted: 15 AUG 2006
- Manuscript Received: 11 MAY 2006
- Cancer Research (16-6) from the Ministry of Health
- Labor and Welfare of Japan
- Scientific Research Expenses for Health
- Labor and Welfare Programs
- Promotion of Cancer Research
- Third Term Comprehensive Control Research for Cancer
- imprint smear;
- lung adenocarcinoma;
- bronchioloalveolar carcinoma component;
Patients with noninvasive, small-sized primary adenocarcinomas of the lung have excellent prognosis after lobectomy. Several researchers have suggested that limited resection could be an acceptable alternative for these patients. Therefore, a preoperative or intraoperative judgment of invasiveness would be one of the critical determinants of the surgical procedure in each case. Cytopathologic findings that can distinguish invasive from noninvasive adenocarcinomas remain to be elucidated.
Imprint smears were obtained from 60 resected adenocarcinomas with nonmucinous bronchioloalveolar features. Thirteen cytologic factors were evaluated: the presence of necrosis, fibrovascular tissue, proportion of macrophages, the presence of large tumor cell clusters, nuclear grooves, nuclear overlapping, variation in nuclear size, chromatin pattern, presence of a nucleolus, intranuclear inclusions, multinucleated cells, spindle cells, and mitosis. Each factor was examined by univariate analysis for correlation with the presence of histopathologic invasion.
In the univariate analysis, 5 cytologic factors—presence of tumor cell clusters consisting of more than 50 tumor cells (P < .001), nuclear overlapping in more than 3 layers (P < .001), presence of nuclear grooves (P = .007), more than 3-fold variation in nuclear size (P < .001), and 1 mitotic cell per 1000 tumor cells (P = .035)—were associated significantly with invasion. Among these, nuclear overlapping in more than 3 layers (P = .003) and more than 3-fold variation in nuclear size (P = .005) were found to be independent predictive factors for invasion by multivariate analysis.
Using imprint smears, the presence of invasion in small-sized primary adenocarcinomas of the lung is predictable by the 2 above-mentioned cytologic findings. Imprint smear cytology may effectively aid intraoperative judgement of invasion in cases where frozen section histology is difficult to interpret. Cancer 2006. © 2006 American Cancer Society.
Recently, small-sized (2 or 3 cm or less) adenocarcinomas of the peripheral lung have been detected more frequently because of improvements in diagnostic radiology,1 particularly CT scan technology. However, some of these tumors are already advanced at the time of diagnosis,2 despite their small size, and they may develop distant metastasis even after complete surgical resection.3 On the other hand, histological analysis has revealed that noninvasive adenocarcinomas are not accompanied by lymph node metastasis, and that patients have excellent prognosis after lobectomy.4, 5 Several researchers have reported that limited lung resection could be an acceptable alternative surgery for these patients. Thus, preoperative or intraoperative information about the invasiveness of each tumor would be important for determining the surgical procedure, i.e. limited resection or lobectomy with lymph node dissection.
In 1999, the WHO/IASLC Classification Panel proposed a revised, stricter definition for bronchioloalveolar carcinoma (BAC), describing it as a noninvasive adenocarcinoma showing pure lepidic growth without invasion of the stroma, pleura, or blood vessels.6 To fulfill this criterion, appropriate tissue sampling is required in cases with a BAC component so that focal areas of invasion can be identified. As a result, a definitive diagnosis of BAC cannot be made from a biopsy specimen. In the previous classifications, mainly because no emphasis was placed on the importance of a lack of invasive growth, widely varying histologic criteria were used in publications describing adenocarcinomas with a BAC component. According to the revised criteria, most lung adenocarcinomas with a BAC component are now included in the category of adenocarcinoma with mixed subtypes. The same proposal was also adopted by the 2004 WHO classification.7
In cytologic specimens, BAC is known to show characteristic features, including a clean background, three-dimensional clusters, flat sheets, papillae, an orderly arrangement of cells with round uniform nuclei, predominance of mucinous cells, overlapping nuclei, irregular nuclear membranes, fine granular chromatin, macronucleoli, intranuclear cytoplasmic inclusions, and nuclear grooves.8–13 Most of these studies stated that it was possible to diagnose BAC using cytologic specimens. However, these studies were performed before the introduction of the 1999 WHO criteria, and therefore it is possible that no distinction may have been made between noninvasive adenocarcinoma (BAC) and invasive adenocarcinoma with a BAC component (AD-BAC).
To clarify the feasibility of cytologic prediction of the invasive component in small-sized lung adenocarcinomas with a BAC component, we analyzed the cytologic features of these tumors in imprint smears.
MATERIALS AND METHODS
Cases and Tumor Specimens
In total, 60 cases of adenocarcinoma of the peripheral lung, measuring 3 cm or less in greatest dimension, with bronchioloalveolar features, without a mucinous component or lymph node metastasis, were selected from the surgical pathology files at the National Cancer Center Hospital between May 2002 and January 2005. No preoperative chemotherapy or radiotherapy had been performed in any of these cases.
Imprint smears were obtained from the maximum fresh cut surface of resected tumor specimens, fixed routinely in 95% ethanol, and then stained using the Papanicolaou technique. The resected specimens were then fixed in an inflated state by transbronchial infusion and/or injection of 10% buffered formalin. All tissue blocks containing tumor tissue were embedded in paraffin. Consecutive 5-μm sections were cut and then stained with hematoxylin and eosin and the elastica van Gieson method.
All histologic specimens were carefully evaluated by 2 or 3 certificated pathologists on the basis of the criteria described in the WHO classification.7 Furthermore, we judged the tumor as AD-BAC when it showed active fibroblastic proliferation4 and/or disruption of the elastic framework mostly in the central part,14, 15 or vascular or pleural invasion.6 As a result, the 60 cases examined in this study were composed of 25 cases of BAC and 35 cases of AD-BAC (Table 1). The research was approved by the Institutional Review Board and an informed consent was obtained from each patient enrolled in the study.
|Factor||BAC (n = 25)||AD-BAC (n = 35)|
|Male/female ratio||8/17 (0.47)||19/16 (1.19)|
|Mean age, y (range)||58 (48–76)||60 (54–74)|
|Mean tumor diameter, mm (range)||10 (4–22)||17 (9–29)|
Imprint smear slides (3 to 4) in each case were examined by 3 observers (NM, YS, and KT) and assessed for the following parameters without knowledge of the clinicopathologic data for each patient.
Thirteen cytologic parameters were selected for evaluation: presence or absence of 1) necrosis, 2) fibrovascular tissue fragments, 3) macrophages outnumbering tumor cells, 4) nuclear grooves, 5) large tumor cell clusters consisting of more than 50 tumor cells, 6) more than 3-fold variation in nuclear size, 7) nuclear overlapping in more than 3 layers, 8) chromatin pattern, and 9) prominence of nucleolus. In addition, the following factors were observed in 10 high-power-view fields (HPF), and judged positive if a specimen showed more than the following: 10) 1 intranuclear inclusion per 100 tumor cells, 11) 1 multinucleated tumor cell per 100 tumor cells, 12) 1 spindle tumor cell, and 13) 1 mitosis per 1000 tumor cells.
Disagreements in judgment were resolved by means of a joint review of the slides using a multiheaded microscope.
Statistical analysis was performed using SPSS 12.0 for Windows (SPSS, Chicago, IL). Chi-squared (χ2) tests and multivariate analysis were used, and a P value of .05 or less was regarded as significant. Each feature was analyzed using univariate analysis. Features that proved to be significant by univariate analysis were then examined by multivariate analysis.
Univariate analysis showed that nuclear overlapping in more than 3 layers (P < .001) (Fig. 1), more than 3-fold variation in nuclear size (P < .001) (Fig. 2), cluster size consisting of more than 50 tumor cells (P < .001) (Fig. 3), presence of nuclear grooves (P = .007) (Fig. 4A), and 1 mitosis per 1000 tumor cells (P = .035) (Fig. 4B) were associated significantly with invasion (Table 2).
|Large tumor cell clusters|
|Fibrovascular tissue fragments|
Multivariate analysis showed that 2 factors—nuclear overlapping in more than 3 layers (P = .003) and more than 3-fold variation in nuclear size (P = .005)—were independent cytologic factors for invasiveness (Table 3).
|Factor||Relative risk||P||95% CI|
|Large tumor cell clusters||0.384||.384||0.032||4.597|
Sensitivity for invasion of nuclear overlapping in more than 3 layers or more than 3-fold variation in nuclear size or both was 89%, 71%, and 69%, respectively; specificity for the same was 80%, 84%, and 96%, respectively (Table 4).
|Factor||Sensitivity, %||Specificity, %|
In the present study, we revealed that 5 cytologic factors—presence of nuclear overlapping in more than 3 layers, more than 3-fold variation in nuclear size, large clusters consisting of more than 50 tumor cells, nuclear grooves, and mitosis—were predictive of invasion. In particular, nuclear overlapping in more than 3 layers and more than 3-fold variation in nuclear size were independent predictive factors for invasion.
Morishita et al.16 examined 8 cytologic parameters for small-sized adenocarcinomas 2 cm or less in diameter. They revealed that nuclear size in BAC and AD-BAC was predominantly small to medium and medium to large, respectively. The variation in nuclear size in BAC and AD-BAC was mild and moderate, respectively. Karyometric analysis revealed that the tumor nucleus area of AD-BAC was larger than that of BAC. It was concluded that BAC can be distinguished easily from advanced adenocarcinoma.
Ohori et al.17 examined 12 cytologic parameters in 6 cases of BAC, 17 cases of AD-BAC, and 12 cases of invasive pulmonary adenocarcinoma without BAC components. They concluded that there was no significant difference in these parameters between BAC and AD-BAC. In the latest WHO classification, the nuclear pleomorphism is described as one of the criteria used in assessing invasion in the histological diagnosis. The variation of nuclear size in the cytologic materials may reflect the above-mentioned histological feature.
These conflicting results may have been due to differences in patient populations or differences in cytologic criteria or both. Furthermore, The cohort of Ohori and colleagues consisted of 6 cases of BAC, 3 of which were mucinous BAC. Mucinous BAC shows a different clinical course18 and radiographic pattern19 from nonmucinous BAC. For this reason, we consider that mucinous BAC should be analyzed separately in consideration of clinical setting. This is why our cohort did not include any cases of mucinous BAC. In addition, the demographics and the tumor size of cases examined in this study were almost equal to those of cases reported previously.
MacDonald and Yazdi20 revealed that diagnosis of BAC was possible in cytologic specimens on the basis of conventional cytologic criteria (monolayered sheets, papillary fronds, dispersed single cells with uniform round nuclei, moderate to abundant and occasionally finely vacuolated cytoplasm, finely granular chromatin, nuclear grooves, intranuclear inclusions, and nucleoli). However, variation in nuclear size as well as nuclear crowding and overlapping were observed in 6 of 40 (15%) cases of histologically diagnosed BAC. Also in the present study, 4 of 25 cases (16%) of BAC showed more than 3-fold variation in nuclear size and 5 of 25 cases (20%) showed nuclear overlapping in more than 3 layers. These results indicate that use of only single cytologic criteria may result in a false-positive diagnosis of invasion. The combination of 2 independent cytologic factors, nuclear overlapping in more than 3 layers and more than 3-fold variation in nuclear size, improved the specificity for invasion. In a clinical setting, we considered that high specificity should be given priority to high sensitivity for diagnostic criteria of invasion. This is because a patient still has a chance of additional surgical resection even if a limited resection has been performed based on a false-negative diagnosis for invasion using cytologic material, whereas oversurgery as a result of false-positive cytology would result in impaired lung function that would never be compensable.
In the histological diagnosis of lung adenocarcinoma with BAC components, the definitive criteria for invasion is still under debate; however, many authors have emphasized the importance of architectural destruction of lung manifested by active fibroblastic proliferation.4 In the present study, we also evaluated the presence of fibrovascular stromal elements and spindle cells. However, fibrovascular tissue fragments in BAC and AD-BAC were present in 18 of 25 (72%) cases and 27 of 35 (77%) cases, respectively. Also, the presence of 1 spindle cell/10 HPF was noted in 5 of 25 (20%) cases and 10 of 35 (29%) cases, and these 2 parameters did not show any significant differences.
Cytologic diagnosis using imprint smears may provide valuable and rapid diagnostic information about invasiveness. Thus, imprint smear cytology may effectively aid intraoperative judgement of invasion in cases where frozen section histology is difficult to interpret. Furthermore, these results provide a theoretical background for preoperative prediction of invasion by bronchopulmonary cytology.
- 6World Health Organization. International Histological Classification of Tumors: Histological Typing of Lung and Plural Tumors. Heidelberg: Springer; 1999., , , et al.
- 7Adenocarcinoma. In: TravisWD, BrambillaE, Muller-HermelinkHK, HarrisCC, eds. Pathology and Genetics: Tumors of the Lung, Pleura, Thymus and Heart. Lyon, France: IARC; 2004: 35–44., , , et al.