Lung tumors were induced in HRAS-driven Multi-Hit mice, which allow stochastic activation of MAPK, RAL, and PI3K effector pathways [46] and LSL-KRAS^G12D mice harboring additional genetic modifications (knockout of RANK, apelin, or ATG5) using inhalation of Cre-expressing viruses. In a subset of HRAS-driven Multi-Hit mice, TP53 or PTEN genes were deleted. In each of the above experiments between 7 and 70 lungs, tumors developed in mouse lungs. A morphologic sequence from hyperplasia to atypical proliferation to adenoma to adenocarcinoma in situ and invasive adenocarcinoma was found.

In all KRAS oncogene activated mice – regardless of additional genetic alterations – hyperplasia of type II pneumocytes could be found (Fig. 22.1), in several experiments also nodular aggregates of pneumocytes (nodular hyperplasia; Fig. 22.2). These proliferations, however, never progressed into high-grade malignancies. In these proliferations, the morphologic differentiation of type II pneumocytes could be seen, sometimes also a cuboidal transformation of pneumocytes, which is well known in human lung pathology as reactive, reparative, or preneoplastic. Nuclear to cytoplasmic ratio was only slightly increased, but not that amount one will see in neoplasia.

KRAS as well as HRAS oncogenes induced another type of proliferation at the bronchioloalveolar junction. This type of proliferation started as an expansion of epithelial cells at the basal layer of the terminal bronchiolar epithelium (basal or reserve/stem cells). The cells were polygonal or cuboidal and had larger than normal nuclei with still regularly distributed chromatin. Nucleoli were inconspicuous; the cells looked similar to regenerating epithelia (Figs. 22.3, 22.4, and, 22.5). In some cases, atypia was seen early on, the polymorphous nuclei were enlarged, the chromatin was coarse, and nucleoli were enlarged (Fig. 22.6). This proliferation later on developed into a micropapillary epithelial proliferation and further on acquired a mesenchymal stalk with newly formed capillary loops and primitive mesenchymal stroma cells, thus presenting as a papillary preneoplasia. This proliferation extended into the peripheral alveoli, completely filling the air spaces (Fig. 22.7). Since the alveoli in mice are smaller compared to human lung, but the mouse transformed epithelial cells are not as small, these lesions often appeared solid (the ratio of cell diameter of pneumocytes II and alveolar diameter is different across the different mammalian species). Only at high magnification, a slit-like alveolar lumen could be seen. Nodules grew larger (up to 1 cm) and developed central necrosis, most probably due to hypoxia (Fig. 22.8). However, no invasion occurred at this time, but the cytological atypia qualified these lesions as an in situ adenocarcinoma: nuclear to cytoplasmic ratio was >50 %, nuclear chromatin was irregular distributed, nucleoli were increased in size, and few mitotic figures were found.

Within these larger nodules of in situ adenocarcinomas, necrosis did occur. Almost concomitant with necrosis, neoangiogenesis started (Fig. 22.9). Primitive stroma cells were seen expressing CD31 and CD34, deposition of newly synthesized matrix proteins occurred, primitive capillaries were formed, and the stroma developed features of a desmoplastic stroma reaction comparable to that seen in human carcinomas (Fig. 22.9). Once desmoplastic stroma has developed, the tumor cells started to invade this stroma, developing into invasive adenocarcinomas (Fig. 22.9). Angioinvasion was found in some cases (Fig. 22.10), but it seems that an interaction of different genes is necessary: in the HRAS Multi-Hit model, angioinvasion was only seen in those adenocarcinomas, which harbored activation of all three RAS effectors (MAPK, RAL, and PI3K) or additional TP53 or PTEN inactivation [48].

In some models, peripheral adenomas developed in addition to nodular hyperplasia. The difference between nodular hyperplasia and adenoma is preservation of the alveolar structure in nodular hyperplasia and the original capillary network confined to the alveolar septa, whereas in adenomas the alveolar architecture is lost, and new capillaries are formed. In all models studied, these adenomas never progressed into a malignant lesion.

Clara cells expressed Clara cell protein 10 (CC10) in normal bronchi and bronchioles. During the papillary proliferations starting from the bronchioloalveolar junction, the expression of CC10 was still retained within the cuboidal cells but, however, was lost in most cells, when adenocarcinomas in situ were formed (Fig. 22.11). Only single cells still expressed this protein. A similar reaction was seen for surfactant apoproteins (SApo): type II pneumocytes expressed SApoA and SApoB, and this expression was retained in pneumocyte hyperplasia, diffuse as well as nodular. In peripheral adenomas, this expression was lost in almost all cells. Few proliferating cells in the papillary bronchioloalveolar junction expressed SApoB. In adenocarcinomas in situ, this expression was lost; however, in invasive adenocarcinomas, SApoB could be demonstrated in the more differentiated cells in the center of the adenocarcinoma, especially in the papillary component. SApoC was expressed in pneumocytes type II in hyperplasia, but single cells also expressed SApoC in the bronchioloalveolar junction papillary proliferations. Also in adenocarcinomas in situ and in invasive adenocarcinomas, single cells and small-cell clusters expressed SApoC.

Using double staining for CC10 and SApoC in selected cases, double staining for these markers occurred in very few or only single cells (Fig. 22.12). Cells coexpressing CC10 and SApoC are regarded as peripheral stem cells residing in the niches of bronchioloalveolar junctions [49]. There was no expansion of these cells indicating that the tumor proliferation most probably started from more differentiated cells and not peripheral stem cells. The proliferation starting from the bronchioloalveolar junction was constantly positive for pan-cytokeratin, ruling out the presence of newly invading hematopoietic stem cells as the source of tumor cells [48].

Within the necrotic centers, newly formed blood vessels could be seen using immunohistochemistry for CD31 and CD 34. When immunohistochemistry was applied for podoplanin, which is a marker of endothelia of small blood vessels and lymphatics, dilated lymphatics were seen at the periphery of the in situ adenocarcinomas, but not in the necrotic centers (Fig. 22.9d, e). Neoangiogenesis of lymphatics was only seen after new blood vessels already appeared in the hemorrhagic or necrotic centers of the carcinomas. Lymphangiogenesis accompanied the invasion of the carcinomas, but did not precede it.

Stromal invasion in all mouse models depended on tumor size, the occurrence of hypoxia in the center, necrosis, desmoplastic stroma reaction, and angiogenesis. Invasion occurred only in large nodules within their center. The morphologic changes preceding invasion were hemorrhage and infarct-like necrosis suggesting a prominent role of hypoxia. Necrosis and hemorrhage were followed by formation of desmoplastic stroma and primitive blood vessels. Carcinoma cells invaded predominantly into the newly formed modified stroma. Importantly, hypoxia seems to promote invasiveness of human carcinomas [50–53] as well indicating that the mouse models could be an ideal tool to study the association of hypoxia, invasion, and metastasis.

Epithelial to mesenchymal transition (EMT) is a mechanism, which facilitates invasion and oriented tumor cell movement within the stroma. It can be easily diagnosed by the spindle cell morphology of the tumor cells, which express cytokeratin, vimentin, and smooth muscle actin (cytokeratin expression can even be lost). In humans, the most instructive lung cancer type displaying features for EMT is sarcomatoid carcinoma (spindle cell and pleomorphic carcinoma). The molecular basis for EMT in lung carcinomas might be pleiotropic and cannot be attributed to single genes [20, 54–66]. In the investigated mouse models, EMT seemed to play a minor role and was only represented in a single tumor by a small focus of cells with less than 1 mm in diameter. Here tumor cells acquired typical spindle cell morphology, but still retained expression of cytokeratin (Fig. 22.13) [48].

Invasion into blood vessel required additional genetic aberrations such as TP53 or PTEN deletion in HRAS Multi-Hit mice. Interestingly none of the adenocarcinomas in various mice displayed metastasis, which might be due to tumor overload. Up to 70 foci in lungs and additional pneumocyte hyperplasia might have reduced the ability of oxygenation, because type II pneumocytes are large and increased the distance between alveolar and capillary lumina. Therefore, even in cases with angioinvasion, the animals might have died due to respiratory insufficiency before metastasis could occur. Moreover, as pointed out above, induction of metastasis might require additional genetic modifications [67, 68].

The morphology of mouse lung carcinomas, induced by exposure to carcinogens, has been described in previous reports [8, 69, 70]. These chemical models never simulated adenocarcinomas of humans although rare cases of mucinous tumors were reported. Even squamous cell carcinomas were predominantly characterized as cystic tumors surrounded by an atypical squamous epithelium without signs of invasiveness. Genetically induced mouse lung tumors [5, 71] resembled human adenocarcinomas more precisely, but the sequence of events in mouse models was most often described by basic researchers not familiar with human lung carcinoma morphology. Recently, a consensus conference established a nomenclature for proliferative lesions evolving in mouse models [72]. This classification is a first step to precisely define the different sequences of proliferation in mouse models, but some aspects are still missing:

Human adenocarcinomas can be separated into the common peripheral types, presenting with a lepidic, acinar, micropapillary, cribriform, and/or solid pattern. Usually one pattern is predominant but rarely only one pattern is present. In addition, peripheral adenocarcinomas can be further separated into mucinous and non-mucinous types. Rarely, a mixed mucinous and non-mucinous morphology can be observed. In addition, some rare variants exist such as fetal-, intestinal-, or colloid-type adenocarcinomas [73, 74]. Moreover, rare central adenocarcinomas with a morphologic pattern resembling bronchial glands but also solid and acinar adenocarcinomas can occur.

Adenocarcinomas in different mouse models morphologically present as papillary (Fig. 22.15) and solid types, where solid forms (Fig. 22.16) have to be carefully evaluated, because they are often pseudo-solid due to the filling of the alveoli with tumor cells. This is a common morphology of in situ adenocarcinomas. Lepidic, micropapillary, cribriform non-mucinous, and various forms of mucinous adenocarcinomas are never observed. Lepidic adenocarcinomas probably could not develop in a mouse lung because of the small diameter of alveoli and the relatively large size of the tumor cells. Therefore, in situ adenocarcinomas often showed a pseudo-solid morphology, but slit-like spaces of alveolar remnants are identified on higher magnification. Other types of adenocarcinomas are not formed because additional stimuli (e.g., for goblet cell formation) are missing during early steps of carcinogenesis [75–79]. Thus, only a small percentage of human pulmonary adenocarcinomas are morphologically represented in the mouse models, and for the vast majority, no suitable model exists.

The histopathology of HRAS- and KRAS-induced tumors suggests that these mice represent a model for non-mucinous peripheral thyroid transcription factor-1 (TTF1)-positive human adenocarcinomas. Papillary and solid differentiation was most commonly found in KRAS models, whereas solid and acinar adenocarcinomas were predominantly seen in HRAS models. Additional genetic modifications might be needed for development of micropapillary or cribriform non-mucinous adenocarcinomas as observed in human patients.

The fate of inhaled material is quite different in humans and mice: particulate materials are predominantly cleared in the upper respiratory tract of mice. The inhaled air circulates within the large sinusoidal areas where particulates are deposited. Therefore, inhaled particulates being either toxic and/or carcinogenic will act primarily in the upper respiratory tract [80, 81]. In addition, the airways of the mouse lung branch in a dichotomous manner, i.e., each bronchus divides into two equally sized bronchial branches. This results in an almost laminar airflow, and particles with an aerodynamic diameter of <2 μm will be deposited in the peripheral lung. In contrast, there is not much clearance of particulates in the nose and nasal sinuses of humans, but due to an asymmetric bronchial branching (a bronchus divides into a main branch with two thirds of the diameter and a minor bronchus with one-third diameter), particulates are deposited in the larger bronchi due to turbulences at the bifurcations. Therefore, even particulates with a small aerodynamic diameter might already be filtered out at central bronchial bifurcations. In human bronchi, the number of branches outnumbers those of mice more than three times, and the cell composition and distribution are different: in human bronchi, the ratio of ciliated to goblet cells is 6–8:1, whereas the number of goblet cells in the mouse is significantly lower. Ciliated cells occur in human lungs down to the larger bronchioles, whereas in mouse lungs, bronchioles are almost devoid of ciliated cells. In contrast, Clara cells in mouse lungs do not only occur in a substantial number in bronchioles, but also larger bronchi, whereas in human lungs, these cells are mainly confined to the terminal bronchioles and alveolar ducts. Secretory cells are forming one compartment in human and mouse bronchioles, but have been renamed as tuft cells or pneumocytes type III in mice. Short microvilli and secretory granules in the cytoplasm characterize these cells [82, 83]. Their function is unknown, but they can be found in well-differentiated peripheral human lung adenocarcinomas. These morphologic differences might partially explain some of the differences between human and mouse adenocarcinomas.