Introduction

Adenocarcinoma is the most prevalent histological type of lung cancer, accounting for 50–70 % of resected lung cancers [1], and its prevalence is still increasing worldwide [2]. Strategies to diagnose and treat this histological type are, therefore, critically important. A new pathological classification for lung adenocarcinoma was proposed by the International Association for the Study of Lung Cancer (IASLC), the American Thoracic Society (ATS) and the European Respiratory Society (ERS) in 2011 [3]. The main revisions from the previous 1999/2004 World Health Organization (WHO) classification [4] were the omission of the term bronchioloalveolar carcinoma (BAC) [5] and abolishment of the category of mixed subtype adenocarcinoma. The term BAC was abolished to avoid confusion between non-mucinous BAC and mucinous BAC, owing to the substantial differences in the biological and clinical features of these subtypes, and due to the need to clearly distinguish them terminologically. In particular, non-mucinous BAC, including the non-invasive early adenocarcinomas, such as type A and type B in Noguchi’s classification [6], shows a remarkably good prognosis, whereas most mucinous BACs are invasive and have a worse prognosis than non-mucinous BAC.

Among other alterations, the mixed subtype adenocarcinoma category was abolished, because it accounted for as many as 80 % of adenocarcinomas. The new classification more precisely categorizes these formerly mixed type adenocarcinomas in accordance with the predominant histological pattern, as determined by the percentages of constituent histological components. Following the revision, preinvasive adenocarcinomas such as Noguchi’s type A and type B have been classified as adenocarcinoma in situ (AIS), while most type C cases are classified as minimally invasive adenocarcinoma (MIA) or lepidic-type invasive adenocarcinoma. Invasive adenocarcinomas are defined by the predominant histological subtype; namely lepidic predominant invasive adenocarcinoma (LPA), acinar predominant invasive adenocarcinoma (APA), papillary predominant invasive adenocarcinoma (PPA), micropapillary predominant invasive adenocarcinoma (MPA), solid predominant invasive adenocarcinoma (SPA) and other variants of invasive adenocarcinoma, including invasive mucinous adenocarcinoma (IMA; formerly termed mucinous BAC, colloid (CA), fetal (FA) and enteric (EA) adenocarcinoma) (Table 1) [7].

Table 1 The new classification of lung adenocarcinoma in resected specimens, 2011 (IASLC/ATS/ERS)
Full size table

Although this reclassification is theoretically reasonable, its clinical impact on the diagnostic, therapeutic, and prognostic considerations requires further validation. Suitable improvements in future revisions warrant broad discussion among researchers, including oncologists, pulmonologists, pathologists, radiologists, molecular biologists and chest surgeons. To date, however, no substantial and comprehensive evaluation of the impact of this classification on chest surgeons has yet appeared.

Here, we broadly review the impact of the new IASLC/ATS/ERS pathological classification of lung adenocarcinoma, with particular regard to chest surgery.

The proportion of adenocarcinoma subtypes among institutions

The proportions of individual adenocarcinoma subtypes obtained using the new classification system are interesting. Reports from different institutions or countries are shown in Table 2 [8–17].

Table 2 The proportion (%) of adenocarcinoma subtypes (IASLC/ATS/ERS) by race
Full size table

The percentages of each subtype differ among countries and ethnicities in several ways. First, countries differ in the proportion of AIS + MIA; the proportions are high in Japanese and Korean patients, ranging from 11.3 to 13.8 %, but are relatively low in China, the US, Australia and European countries, ranging from 0 to 5.1 % of the cases. A similar trend was seen for LPA, with 8.1–28.3 % of cases in Japan and Korea being LPA vs. 2.7–10.6 % of cases in other regions. It is unclear whether the low proportions of lepidic-type adenocarcinomas (AIS + MIA + LPA) [18] in areas other than Japan and Korea are due to racial and ethnic causes, or whether some other reason exists. One plausible reason is the prevalence of chest computed tomography (CT) in these regions. This is because chest radiography (CR) alone is generally unable to image these lesions, which show up as focal ground-glass opacities, which are generally detectable only on chest CT. In Japan, chest CT plays an important role in detecting lepidic-type early adenocarcinomas. Increased proportions of AIS, MIA and LPA in resected lung cancers improve the postoperative survival in this patient group. In fact, the postsurgical survival in patients with stage IA non-small cell lung cancer (NSCLC) is significantly better in Japan than in the US [19], although the reason for this has not been identified.

Second, the most frequent subtype appears to differ by race, with APA being the most prevalent in Caucasians, ranging from 40 to 46.9 %, whereas PPA is the most prevalent in Asians, ranging from 30.7 to 40.7 %, although exceptions exist. As with the frequency of lepidic-type adenocarcinoma, it is unclear whether the differences in the most prevalent subtype do in fact reflect differences among countries or ethnicities. Another plausible reason is the variation in the interpretation of pathological diagnoses between countries or institutions, because APA and PPA co-exist in many adenocarcinoma cases. Further, variations might be present in the criteria used to interpret APA and PPA among pathologists in different regions or institutions, potentially biasing the subtyping toward the predominant histological type. However, since most studies have found no major survival difference between APA and PPA [8, 9, 12], an inconsistent classification between these two subtypes might not affect the postoperative survival.

Third, the percentage of SPA varied widely among reports, from 6.5 to 37.6 %. It has been noted that large cell carcinoma lacking squamous differentiation is clinicopathologically indistinguishable from SPA [20]. Additionally, large cell carcinoma and SPA are similar with regard to the KRAS mutation rates, at 43 and 38 %, respectively (p = 0.62). Pathologists should, therefore, carefully distinguish large cell carcinoma or other undifferentiated carcinomas from SPA.

Finally, the percentage of MPA also varied widely, ranging from 1.0 to 16.1 %. Given the poor prognosis of this subtype [21, 22], different proportions of MPA in patient groups would be expected to markedly affect the postoperative survival rates. Solving these problems in the pathological diagnosis of subtypes requires the establishment of a common consensus for objective criteria [23]. Achieving such a consensus would be aided by international or multi-institutional workshops for pathologists to standardize the pathological diagnosis of lung adenocarcinoma.

Radiographic findings of individual subtypes

It is well recognized that atypical adenomatous hyperplasia (AAH) and lepidic-type adenocarcinomas, including AIS, MIA and LPA, usually show up as focal ground-glass opacities (GGOs) on chest CT. However, only a few studies [15, 24] have evaluated the features of chest CT findings for the differential diagnoses of other adenocarcinoma subtypes under the new classification. In a retrospective study [24] of 300 resected small lung adenocarcinoma lesions less than 20 mm in diameter, 114 of the 142 air-containing lesions were AIS and 28 were MIA. In contrast, 30 of the 158 solid-density lesions were AIS, 24 were MIA and 104 were invasive adenocarcinoma. Invasive adenocarcinomas commonly showed notches and pleural tags. Thus, the study did not show complete correspondence between the radiographic findings and histological subtype. In a second study [15], adenocarcinoma subtypes shared similar frequencies of CT features, such as bubble-like lucency, cavities, notching and a lobulated border. The frequencies differed only for an air bronchogram and round shape; an air bronchogram was found in 100 % of IMA and a round shape was found in 62.5 % of AIS cases. These two studies indicate that it is difficult to distinguish individual subtypes based on the radiological findings, particularly for tumors forming sub-solid or solid lung nodules such as APA, PPA, SPA and MPA.

An elevated maximum standardized uptake value (SUVmax) in fluorodeoxy glucose-positron emission tomography combined with CT (FDG-PET/CT) was found to be closely associated with SPA or MPA, and patients with these adenocarcinoma subtypes had a higher risk of recurrence than patients with APA or PPA in an intermediate-risk group [25]. This study indicates the usefulness of FDG-PET/CT combined with subtyping for predicting the prognoses of patients with adenocarcinomas.

Postoperative survival of patients with the different subtypes

The postoperative survival of the patients with each subtype in different institutions is shown in Table 3. Although there were differences in the stage of disease, common characteristics of each subtype can be recognized.

Table 3 The postoperative survival according to the lung adenocarcinoma subtype
Full size table

Adenocarcinoma in situ and minimally invasive adenocarcinoma

Adenocarcinoma in situ and MIA are generally detected as small round-shaped GGO lesions with a relatively clear margin on high-resolution chest CT [26]. When a solid component cannot be seen in the shadow, these lesions are sometimes called pure GGOs, most of which are AIS. Multiple studies [27–29] have reported that AIS and MIA are lepidic-type early adenocarcinomas which show no recurrence after complete resection, and the postoperative 5-year disease-free survival (DFS) for these subtypes has been 100 % [8, 9, 12, 16, 17].

Lepidic predominant invasive adenocarcinoma

Non-invasive and invasive lepidic-type adenocarcinomas consist of three subtypes: AIS, MIA and LPA [18]. In the probable carcinogenesis sequence from AAH to invasive adenocarcinoma [30], LPA is postulated to be the stage following MIA. The postoperative 5-year DFS in patients with LPA has ranged from 71.9 to 93.8 %, and the 5-year overall survival (OS) ranged from 86 to 100 % (Table 3) [8, 9, 12, 16, 17]. In an analysis [18] of resected specimens from 139 patients with lepidic-type adenocarcinoma (two with AIS, 34 MIA and 103 with LPA), no recurrence was found in patients with AIS and MIA. In contrast, patients with LPA experienced a recurrence rate of 8 %, which appears to be lower than the 19 % reported for non-lepidic-type adenocarcinomas (p = 0.003).

Acinar predominant invasive adenocarcinoma

The 5-year DFS rates after the resection of APA have ranged from 54 to 84 %, and the 5-year OS rates ranged from 67 to 81.2 %. These rates, therefore, appear worse than those for LPA (Table 3). Of note, the prognosis in patients this subtype was worse in those who had tumors with a cribriform component [31]. This specific type was a distinct subtype with a high risk of recurrence, and the presence of a cribriform pattern is an independent predictor of recurrence which identifies a poor prognostic subset among APA. In fact, the 5-year DFS rate for patients with a <10 % cribriform pattern was 84 %, whereas that for patients with a ≥10 % cribriform pattern was 73 % (p < 0.001). The new classification does not refer to a cribriform component in APA, but this type might warrant independent classification in a further revision.

Papillary predominant invasive adenocarcinoma

The 5-year DFS rates after the resection of PPA have ranged from 56.1 to 83 %, and the 5-year OS rates ranged from 71 to 74.7 %. These survival rates were similar to those of APA, but were apparently worse than those of LPA (Table 3). This subtype is the most common subtype described in most reports from Asian countries to date.

Micropapillary predominant invasive adenocarcinoma

The 5-year DFS rates after the resection of MPA have ranged from 0 to 67 %, and the 5-year OS rates ranged from 38 to 62 %. These survival rates are worse than those for APA or PPA (Table 3). Other studies have supported these findings. For instance, patients with micropapillary and solid predominant tumors had a significantly worse DFS than those with other subtypes (p < 0.001) [13]. The mean OS times were shown to differ significantly among LPA (78.5 months), APA (67.3 months), SPA (58.1 months), PPA (48.9 months) and MPA (44.9 months) (p = 0.007) [10]. This latter study is unique, because it found that the OS was worse for PPA than for SPA.

Solid predominant invasive adenocarcinoma

The 5-year DFS rates after the resection of SPA have ranged from 43.3 to 70 %, and the 5-year OS rates ranged from 39 to 58 %. These rates are worse than those of APA and PPA, and were similar to those of MPA (Table 3).

Variants of invasive adenocarcinoma

This group includes rare subtypes, although IMA is the most common. In patients with IMA, the 5-year DFS rates have ranged from 76 to 88.8 %, and the 5-year OS rates have ranged from 51 to 88.8 %. These rates are superior or at least comparable to those of APA and PPA (Table 3). The other variants, namely CA, FA and EA, are rarely seen, and survival information is not available.

Subgroups of aggregated subtypes

Although the results of several studies [32] did not reach statistical significance in either univariate or multivariate analyses, probably due to the small number of patients with these individual subtypes, most studies [9, 10, 12, 13, 17] demonstrated the predictive value of the new adenocarcinoma classification. When the patients with these subtypes were grouped into low-, intermediate- and high-risk groups, postoperative survival differences among the aggregated groups could be clearly detected (Table 4).

Table 4 Postoperative survival in the aggregated lung adenocarcinoma subtype groups
Full size table

Associations of driver mutations with subtypes

Various driver mutations in lung adenocarcinomas have been reported. The genes that have received the most study to date include the epidermal growth factor receptor (EGFR) gene, the Kirsten rat sarcoma viral oncogene homolog (KRAS) gene and the echinoderm microtubule-associated protein-like 4-anaplastic lymphoma receptor tyrosine kinase (EML4-ALK) fusion gene [33–37]. Other aberrant genes rarely found in lung adenocarcinomas are V-raf murine sarcoma viral oncogene homolog B1 (BRAF), phosphatidylinositol-4,5-biphosphate 3-kinase, catalytic subunit α (PIK3CA), human epidermal growth factor receptor type 2 (HER2), mesenchymal–epithelial transition factor tyrosine kinase (MET) and mitogen-activated protein kinase kinase 1 (MEK1) [38–41]. Small molecule inhibitors of tyrosine kinases involved in the signal transduction for cell proliferation have been developed for clinical use, including gefitinib, erlotinib and afatinib for EGFR, and crizotinib and ceritinib for EML4-ALK [33, 34, 42–46].

Several studies [11, 15–17, 47–51] have analyzed the relationships between adenocarcinoma subtypes and the well-documented genetic mutations in KRAS, EGFR, and ALK-EML4 (Table 5). KRAS-mutated adenocarcinomas and EGFR-mutated adenocarcinomas differ in several clinical features. A meta-analysis [52] suggested that KRAS mutations are associated with a worse OS in patients with NSCLC, particularly in patients with adenocarcinoma in the early stages. In contrast, several other studies [53] showed that patients with completely resected NSCLC carrying the EGFR mutation have a significant survival advantage over patients with tumors harboring wild-type EGFR. A Japanese study [54] comparing the postoperative survival in 32 adenocarcinoma patients with the KRAS mutation and 148 patients with the EGFR mutation showed a better OS in the EGFR-mutated group (p = 0.0271).

Table 5 Percentages of driver mutations in the lung adenocarcinoma subtypes
Full size table

In terms of the survival after tumor recurrence, the EGFR-mutated group had a better median survival time (46.7 months) than the KRAS-mutated group (26.0 months). In a US study [55] which analyzed an adenocarcinoma group, 25 % of whom had a KRAS mutation and 20 % had an EGFR mutation, the patients with EGFR-mutated lung cancers were at lower risk of death than those without EGFR mutations (p < 0.001). When smokers and non-smokers were compared in Asian lung cancer patients, the EGFR mutation rate was found to have decreased, while the KRAS mutation rate increased with an increased cumulative smoking dose, measured in pack years (no. of packs/day × years of smoking) [48]: the respective mutation rates for <10, 11–20, 21–30, 31–40, 41–50 and >50 pack years were 74.2, 61.5, 39.7, 39.1, 30.4 and 15.1 % for EGFR, but 9.7, 7.6, 15.5, 19.6, 21.7 and 27.3 % for KRAS. Based on these findings, an EGFR mutation is considered to be a tobacco-unrelated good prognostic factor frequently seen in Asians, a KRAS mutation is considered to be a tobacco-related worse prognostic factor frequently seen in Caucasians.

The studies listed in Table 5 illustrate the following important points. First, the somatic mutations in lung adenocarcinoma differ by race or ethnicity. The frequency of KRAS mutations is high in Caucasians, whereas the frequency of EGFR mutations is high in Asians, regardless of the adenocarcinoma subtype. Concerning EGFR mutations and adenocarcinoma subtypes, the trends in mutation rates in PPA differed between Caucasians and Asians, being relatively high (36.1–68.5 %) in Asians and low (0–13.7 %) in Caucasians. In contrast, a KRAS mutation was commonly found at a high frequency in IMA cases in both Caucasians and Asians. The EGFR mutation rates in SPA were low (2.8–15.8 %) in both Caucasians and Asians [49]. Moreover, one study [56] reported an overall response rate for EGFR-TKI in EGFR-mutated SPA which was significantly worse than in other subtypes (61 vs. 88 %, p = 0.03), and the median progression-free survival and OS after EGFR-TKI treatment were significantly shorter for patients with SPA than for those with other subtypes of lung adenocarcinoma, suggesting that EGFR-TKI is less effective for SPA harboring an EGFR mutation. Patients undergoing EGFR-TKI therapy should, therefore, be treated after considering the adenocarcinoma subtype.

With regard to the histological features of lung cancer harboring the ALK-EML4 fusion gene, ALK-rearranged tumors were associated with a younger age, frequent nodal metastasis and a higher stage of disease at diagnosis [47]. In addition, these ALK-rearranged tumors were more likely to be found in SPA (56.4 %), followed by IMA (30.0 %) and PPA (27.9 %), and were not observed in LPA (Table 5). In a multivariate analysis, the most significant morphological features that distinguished ALK-rearranged tumors from ALK-negative tumors were cribriform formation, the presence of mucin-containing cells, a close relationship to adjacent bronchioles, the presence of psammoma bodies and the SPA subtype [47]. However, another study [17] showed the conflicting result that ALK-positive tumors were more frequently found in MPA (15.0 %) and APA (14.4 %) cases than SPA (6.5 %) cases. The mutation rates among adenocarcinomas in 349 female Chinese never-smokers were 76.2 % in EGFR, 4.3 % in EML4-ALK fusions, 4.6 % in HER2, 2.0 % in KRAS and 0.6 % in BRAF [57].

Future issues related to surgical treatments and adenocarcinoma subtypes

Subtypes with a better prognosis

Pure GGO lesions only detectable on chest CT, most of which belong to type A and type B in the Noguchi classification [6], can be cured by sublobar resection, such as wedge resection or segmentectomy [27, 28]. Since the new criteria have recategorized localized non- and slightly invasive small BACs as AIS and MIA, these two subtypes can be resected completely by limited resection without systematic nodal dissection, as described in a review article on the new adenocarcinoma subtypes [3]. A preoperative definite diagnosis of these subtypes by transbronchial biopsy or percutaneous needle biopsy is difficult, and a pathological diagnosis is usually obtained using the resected specimen after surgery. Thus, the most important preoperative information for the diagnosis of AIS and MIA is image analysis by high-resolution chest CT. AIS usually grows slowly, but the speed differs on a case-by-case basis, and a consensus on the optimum timing of the resection for AIS showing pure GGO has yet to be obtained.

Recently, cases showing multifocal AIS or MIA have been reported [58]. For lesions located in the periphery of the lung, multiple wedge resection for early lung cancers is a reasonable means of preserving the lung function. However, if multiple lesions are located near the hilum in different lobes, organ-sparing surgery is often difficult. In such cases, the timing of surgery and selection of surgical procedure should be done following a thorough evaluation of the clinical factors, including the patient age, speed of tumor growth and estimated postoperative residual lung function.

Lepidic predominant invasive adenocarcinoma is more often found in females, and lymph node metastasis is less common for LPA than for other non-lepidic subtypes [59]. In one study [60] which analyzed adenocarcinomas less than 20 mm in diameter, the 5-year DFS rate in 21 patients with LPA was 100 %, compared with 74 % in 43 patients with non-lepidic invasive adenocarcinomas (p = 0.035). Since the 5-year DFS and OS after the resection of LPA were around 90 % in most reports, limited resection and selected nodal dissection might be applicable for this subtype. A recent randomized trial which compared lymph node sampling with systematic nodal dissection for stage T1-2N0 nonhilar N1 NSCLC showed that systematic nodal dissection identified occult node-positive disease in 3.8 % of patients, but was not associated with a benefit in terms of the overall survival [61]. These results may support the omission of, or selective use of, mediastinal nodal dissection [62] for this less-invasive adenocarcinoma subtype, LPA. However, the presence of a residual tumor at the surgical margin should be carefully confirmed during surgery to prevent local recurrence, particularly when the tumor is large [18]. Even in small tumors showing GGO less than 2 cm in size, adenocarcinoma developed as cut-end recurrence in four of 26 patients who underwent limited resection more than 5 years after initial surgery [63]. The authors of that study concluded that limited resection should still be performed only in a trial setting, even for GGO lesions. Ongoing randomized trial comparing segmentectomy vs. lobectomy for small-sized peripheral NSCLC will provide useful information about the efficacy of limited resection [64].

Subtypes with a worse prognosis

The percentage of the micropapillary component was reported to be associated with local recurrence after limited lung resection: lung adenocarcinoma with a micropapillary component of 5 % or greater was associated with a greater risk of recurrence than that with a micropapillary component of less than 5 % (5-year cumulative incidence of recurrence, 34.2 vs. 12.4 %, p < 0.001) [21]. This report directly indicated that there is a histological characteristic of adenocarcinoma which rendered affected tumors unsuitable for limited resection. Since most adenocarcinoma subtypes can be pathologically diagnosed only by the resected specimen after surgery, the problem is, therefore, how to diagnose the adenocarcinoma subtypes preoperatively to enable selection of the proper surgical option. We emphasize the importance of a multidisciplinary tissue management strategy to obtain a preoperative diagnosis for tumor subtyping from small biopsy samples [65].

As the postoperative survival of MPA and SPA is significantly worse than that of other subtypes [10, 13, 66], combined therapy with surgery requires further discussion. MPA and SPA are common in tumors greater than stage I, and a size ≥2.5 cm, pure solid type and a SUVmax ≥ 7 are all predictors of a poor DFS [22].

A patient group with advanced-stage MPA + PPA + SPA showed a better response rate, progression-free survival and OS to platinum-based chemotherapy than an LPA + APA group (36.9 vs. 25.4 %, p = 0.034; 6.4 vs. 5.5 months, p = 0.009, and 25 vs. 16.8 months, p = 0.023; respectively) [67]. This study concluded that patients with high-risk adenocarcinoma in advanced stages have a longer OS than patients with intermediate-risk adenocarcinoma, unlike the case in early stages, probably due to a better response to chemotherapy. If this is true, the ability of radical surgery with systematic lymph node dissection combined with induction or adjuvant chemotherapy to improve the survival of SPA or MPA should be confirmed in future. Although their prognosis is generally worse than that of other subtypes, MPA and SPA show differences in genetic aberrations: the EGFR mutation rate is higher in MPA than SPA, whereas KRAS mutations are more common in SPA. Therapeutic strategies, including the use of EGFR-TKI, should therefore be differentially considered for MPA and SPA.

Future directions

The strategies used to treat lung adenocarcinoma should be planned after considering the new IASLC/ATS/ERS classification for adenocarcinoma subtypes. Particularly for chest surgeons, the important considerations in surgery for lung adenocarcinoma include the extent of lung resection, omission of nodal dissection and appropriateness of combined therapy [68–70]. After the new classification has been validated, additional knowledge will allow the subtyping to be further improved and refined, and for a consensus to be reached regarding the optimum treatment of each subtype.