Patient Characteristics and Histopathological features
Demographic data of patients were summarized in Table 1. Patient ages ranged from 23 to 83 years (median 58 years). There were 96 patients at stage I (36.7%), 21 at stage II (8.1%), 79 at stage IIIA (30.3%) and 65 at stage IIIB and IV (24.9%).
All patients were diagnosed of invasive LACs. 114 (43.7%) had uniform pattern and 147 (56.3%) harbored at least two types of constituents. 154 LACs (59%) had EGFR mutation, including acinar (55.1%), lepidic (85.7%), micropapillary (63.6%), mucinous (9.1%), papillary (81.8%) and solid (40.7%). EGFR statuses in histological subtypes were significantly different and mutations were more prevalent in lepidic, micropapillary and papillary LACs (P<0.01)。17 LACs (6.5%) were detected with ALK fusion gene. The median PFS time of 261 patients was 17.6 months (range 0.3–72 months).
PD-L1 expression varied in LAC histological pattern and with tumor stage
PD-L1 expression varied in LAC (Fig 1). PD-L1 was noticed diffusely expressed in some types of tumor cells (Figure 1A). However, patchy or heterogeneous expression was also detected in a same tumor mass (Figure 1B), and even within the same component (Figure 1C and D). Therefore, we compared PD-L1 expression level in different histological patterns (Fig 2). PD-L1 TPS in LACs with solid component was higher than that in LACs without this pattern (P=0.0023). LACs with mucinous components expressed in lower PD-L1 level compared to those without this pattern. Similarly, LACs harboring papillary pattern had slightly lower level of PD-L1 TPS than LACs without it (P=0.051). Acinar and micropapillary constituents were not significantly associated with PD-L1 expression level.
We also assessed PD-L1 expression by using three tiered grading system: 1%, 1%-49% and >= 50% (Table 2). We assessed PD-L1 TPS in predominant components and found that histological types were associated with PD-L1 expression. In the group of PD-L1 TPS >=50%, Mucinous (0) and lepidic (1.6%) LACs were generally short of high-level of PD-L1 expression, whereas solid (41.3%) had the highest PD-L1 TPS (P<0.01). Likewise, solid LAC was the type inclined to overexpress PD-L1 (41.3%, P<0.01). Mixed LACs with the level of PD-L1 TPS >=50% were prevalent in LACs with two components than those with more than two components (P=0.04). 144 patients were diagnosed at advanced stage (III and IV). We noticed that more than 50% cancer cells expressed PD-L1 in 43 LACs (68.3%) at advanced stage versus those (31.7%) at earlier stage (P=0.04). PD-L1 TPS were not significantly associated with EGFR status (P=0.7). In our study, 58.8%, 29.4% and 11.8% in ALK positive lung adenocarcinomas (N=17) were found PD-L1 expression tiered via cutoffs of <1%, 1%-49% and >=50% tumor cells (P > 0.05).
PFS time of LAC patients with PD-L1 expression and histological subtypes
We firstly assessed PFS time of group <1%, 1%-49% and >=50%. Of note, we found that high PD-L1 TPS (>=50%) was associated with shorter PFS (median 15.3 versus 22.9 months, p < 0.05) in the entire cohort (Fig. 3A). To address whether the results were confounded by histological components, we observed the PFS time in different histological subtypes and found that PFS time was not significantly longer in three PD-L1 TPS groups among LACs with single histological pattern (P>0.05) (Fig 3B). PFS time was markedly longer in LACs with PD-L1 TPS <1% (median 25.5 months) and harboring two histological components versus those with PD-L1 TPS >=50% (median 13.6 months, P<0.05) (Fig 3C). LACs with >=2 histological components didn’t have significant difference in PFS time (P>0.05) (Fig. 3D). Component-dependant heterogeneous PD-L1 expression were compared in LACs with such components as acinar, solid, micropapillary and papillary patterns. LACs with predominant lepidic components had longer PFS time (median 55.2 months) than other subtypes (P<0.001) (Fig 4A). Among LACs with acinar pattern, the median PFS of the group with PD-L1 TPS <1% was significantly longer than that with PD-L1 TPS >=50% (P=0.031) (Fig 4B). The median PFS (17.8 months) of group PD-L1 TPS < 1% among LACs with solid pattern was slightly longer than that (9.3 months) of PD-L1 TPS 1%-49% group (P=0.055) (Fig 4C). The median PFS time of groups PD-L1 TPS < 1% in LACs with micropapillary components was noticed, even though not significantly, longer than those with higher TPS (P>0.05) (Fig 4D). Similarly, the median PFS of groups PD-L1 TPS <1%, was not significant difference with other groups (P>0.05) (Fig 4E).
Intratumoral Heterogeneous EGFR activating/naïve resistant mutations and different abundance in coaltered LACs
We used amplification refraction mutation system and digital droplet PCR to analyze EGFR hotspot mutation. Eighteen patients had both EGFR sensitive (E19del/L858R) and naïve resistant mutation (T790M). We also observed their histological features, mutant EGFR abundance and expression level of PD-L1 to reveal the biological divergence of these components. All samples were further microdissected to determine the abundance of EGFR mutation in each histological component. We found that T790M mutation were generally accompanied with E19del/L858R except only one LAC harboring single T790M mutation (Table 3). Naïve resistant mutation predominantly coexisted with sensitizing mutations in micropapillary, papillary and lepidic components. Acinar, though, accreting with other components more frequently exhibited EGFR wild-typed in tumor parenchyma. As shown in Figure 5, cancer cells captured in area I and II were of the acinar and solid subtype, which were negative for both sensitizing and resistant EGFR mutation, whereas cells in area III were positive for both sensitizing and resistant mutations. Of interest, the relative abundances of sensitizing and resistant mutations in the same dissected tumor focus varied irregularly. We observed that sensitizing/T790M mutation occurred predominantly in micropapillary and papillary components, whereas absent in acinar and solid components (Fig 5A and 5B). In other words, tumor cells in same AC pattern from a same dissected region harbored heterogeneous cell population (Fig 5C, 5D and 5E). These characteristics might be ignored by mass-based detection. The findings prompted that some tumor cells from this area were heterogeneous in their biological nature and might had inconsistent response to corresponding therapy.
Intratumoral Heterogeneous EGFR status and expression of PD-L1 in LACs
Among the eighteen T790M-positive LACs, 10 LACs had acinar pattern and 6 of them were positive with PD-L1. For these 18 patients, each pattern had been captured and detected by ddPCR to confirm the intratumoral T790M abundance. Nine LACs consisting of more than one pattern had heterogeneous EGFR status and EGFR wild-typed components were positive for PD-L1 expression. We noticed that acinar component was apt to express PD-L1 in a tumor mass and micropapillary and papillary more frequently had both sensitizing and T790M mutation, meanwhile, less often overexpressed PD-L1 protein. We found that PD-L1 expression occurred in acinar and solid areas, but not in micropapillary area (Fig. 5). Cancer cells from different areas in a tumor mass possessed different EGFR statuses and PD-L1 expression. As shown in Figure 5C-E, both areas I and II from one tumor were negative both for E19del and T790M. However, area III from the same tumor coaltered with both E19del and T790M with different abundance. Similar results were observed in other tumors (Table 3). Eight of 10 EGFR-mutated LACs lacking PD-L1 protein were found with single components. However, EGFR mutated components in composite LACs, although in part, still expressed PD-L1 (Table 3). Discordant EGFR status and PD-L1 expression suggested that a tumor mass harbored genetic aberration.