The fibre paradigm identifies the geometry of fibres as their most important toxicological characteristic and not the chemical make-up, except in so far as the composition makes a contribution to biopersistence (see later). This independence from composition is evident in the fact that the paradigm embraces fibres composed of diverse materials including amphibole and serpentine asbestos minerals, vitreous and ceramic fibres and an organic fibre (reviewed in [12]). Diameter is important because of the central role that fibre diameter plays in defining aerodynamic diameter (D_ae) and the dependence of pulmonary deposition on D_ae [13]. Clearance from beyond the ciliated airways is dominated by slow, macrophage-mediated clearance [14] and so fibres which deposit there have the potential to contribute most to build-up of dose. Length impacts little on D_ae for thin fibres [15] except when length is sufficient to cause interception, a mechanism of particle deposition that is confined to fibres, involving the centre of gravity of the fibre following the airstream at a bifurcation whilst the tip of the fibre makes contact with the wall, resulting in deposition. The penetration of long fibres (>50 μm) beyond the ciliated airways is explicable on the basis that the aerodynamic diameter of a straight fibre is around 3 times its actual diameter [15]. This results from its alignment with the airflow as the fibres move aerodynamically through these tubes, aligned along the axis of the bronchial tree.

The evidence demonstrating that length is a key factor in pathogenicity of fibres comes from a number of sources but the best data are from experimental toxicological studies where it is possible to isolate length categories and assess their effects, unlike the mixed nature of human exposures. In the seventies Stanton carried out a large number of studies aimed at understanding the role of fibre characteristics in mesothelioma using implantation of fibres in gelatin, directly onto the pleural mesothelial surface. Although this is a highly artificial exposure, in a summary of these studies [16] Stanton identified that carcinogenicity was related to 'durable' fibres longer than 10 μm. In the study by Davis et al. [17] rats were exposed in a chamber to clouds with equal airborne mass concentration of either long amosite asbestos fibre or a short fibre amosite sample obtained from it by ball-milling. After lifetime exposure there was substantial tumour and fibrosis response in those rats exposed to the long amosite and but virtually no response in rats exposed to the short amosite. Adamson et al. used long and short crocidolite and following deposition in mouse lungs reported fibrosis [18] and proliferative responses [19] at the pleura with the long, but not the short samples. The mouse peritoneal cavity has been used as a model of direct mesothelial exposure and much greater toxic [20], inflammatory [21] and granuloma-generating [4] responses were evident in mice that were exposed to high doses of long fibres than was seen with shorter fibres. In vitro systems have also demonstrated the greater potency of long compared to short fibres in assays of pro-inflammatory and genotoxic activity [22-27].

Kane and co-workers [3] noted that long asbestos fibres accumulated preferentially at the peritoneal face of the diaphragm around the stomata since they could not be cleared through them due to their length. Kane et al. contended that retention of long fibres at the diaphragmatic mesothelial surface initiated inflammation, proliferation and granuloma formation. Short fibres did not cause this effect, easily exiting through the stomata. However, if short fibre were injected at such high dose that their sheer volume blocked the stomata this prevented clearance allowing retention, resulting in inflammation [3]. We confirmed that low dose exposure of the mouse peritoneal cavity to long multi walled carbon nanotubes (MWCNT) [34] resulted in accumulation of the long CNT at the diaphragm, suggesting that they are also too long or bulky to exit through the stomata. Retention of the long CNT in the peritoneal cavity initiated granulomas with classical foreign body giant cells in the peritoneal lavage (Figure ​(Figure5)5) and in the granulomas (Figure ​(Figure6)6) [2]. Short, tangled CNT were not retained and were never seen at the diaphragm or viscera and did not induce inflammation or granuloma formation, their absence from sections strongly suggesting that were cleared through the stomata.

Therefore, we would postulate that there are two important parts to the mechanism of the pro-inflammatory effects of long fibres in the peritoneal cavity, shared by both asbestos and long MWCNT

i) failure of long fibres to negotiate the diaphragmatic stomata with subsequent retention of the long fibre dose at the diaphragm; this contrasts with smaller particles which easily leave the peritoneal cavity through the diaphragmatic stomata to accumulate in the parathymic nodes [33],

ii) at the point where the long fibre dose accumulates at the peritoneal face of the diaphragm macrophages attempt to phagocytose the long fibres; they then undergo frustrated phagocytosis stimulating inflammation and mesothelial cell damage, leading to chronic inflammation and granuloma development [2].

The consequences of pro-inflammatory and fibrogenic effect of long fibre retention at the diaphragm were most evident in the extent of the granuloma/fibrosis response seen 6 months following instillation of 10 μg of long or short nanotubes (see [2] for a full description of the NT tang 2 and NT long 1 nanotubes used in this study). As Figure ​Figure66 shows, the presence of quite a large aggregate of short/tangled nanotubes produced very little tissue reaction (Figure ​(Figure6A6A and ​and6B).6B). In contrast, loose aggregates and singlet fibres of long nanotubes caused a florid granuloma response (Figure ​(Figure6C6C and ​and6D).6D). The multi-layered basketwork -like arrangement of largely acellular collagen in the granuloma is strongly reminiscent of the structure of an asbestos-induced pleural plaque (see later).

The visceral and parietal mesothelium are both composed of a single layer of mesothelial cells, a basal lamina of connective tissue and a loose connective tissue layer with blood and lymph vessels. Mesothelial cells have several functions in normal pleural action [35]. Pleural fluid is constantly produced by hydrostatic pressure from the sub-pleural capillaries [35], supplemented by glycoproteins secreted by the mesothelial cells [36]. The pleural fluid and its constant outflow (see below) maintains tight coupling of the lungs to the chest wall, allowing diaphragmatic muscle contraction and relaxation to expand and passively relax the lungs during breathing movements. The pleural space is a narrow, variable space, that is up to 20 μm or so in sheep rapidly fixed at death and presumed to be similar in size in humans [35]. The pleural fluid turns over rapidly [37], continuously exiting through stomata in the parietal (not the visceral) pleura via lymphatic capillaries; these stomatal openings on the parietal pleural surface are between 3 and 10 μm in diameter (Figure ​(Figure8).8). They are often found in association with 'milky spots', large accumulations of leukocytes present on the parietal pleura [38] and presumed to be involved in immune activity in the pleural space. The pleural fluid outflow through these stomata drains to the lung lymph nodes in the region of the mediastinum and this pathway is important in the clearance of particles and fibres that reach the pleural space (see below). The drainage of fluid from the pleural space carries particles in the lymph to the hilar lymph nodes, mediastinal lymph nodes, parasternal lymph nodes and posterior mediastinal lymphoid tissue [39]. The stomata are most densely situated in the most caudal and posterior intercostal spaces although and are more lightly scattered in more cranial and anterior intercostal regions [40].

Evidence that all particles pass through the pleura comes from a substantial literature concerning the almost universal existence of these "black spots", observable on the parietal pleural wall at autopsy. These mark the stomata and arise where particles must focus to exit the pleural space at the stomata and where they enter the sub-mesothelial connective tissue around the stomatal mouths. In the study by Mitchev et al. [41], 150 consecutive necropsies of urban dwellers were examined in Belgium. Of the 96 male and 54 female necropsies, whose age ranged from 22-93 years, black spots were almost invariably seen (>90% of autopsies) on the parietal pleura. The authors noted that their location appeared to be related to the structures responsible for the lymphatic drainage of the pleural cavity and they considered these to mark the points of pleural fluid resorption. Black spots were also present on the pleural face of the diaphragm, suggesting that there is pleural fluid outflow in a caudal direction. The black spots in the Mitchev study of normal individuals at autopsy clearly reflects that deposited soot particles normally pass through the pleura some of them accumulating in the parietal pleural wall forming black spots. The black spots contain particles and elicit a tissue response which is a low grade in city dwellers, where there is accumulation of dust-laden macrophages and lymphocytes. In coal miners, however, with their large exposures to particles, the mixed dust particles trigger a higher-grade inflammatory reaction of the parietal pleura with concomitant low grade fibrosis in the 'black spots' which are very pronounced [42]. Occasionally pleural inflammatory reactions to interstitialisation of the mixed dust at the black spots are more pronounced, producing more severe granulomatous structures with concentrically arranged collagen fibres [42]. In one study [42], 12 patients with black spots (8 at autopsy and 4 surgically) who were largely miners, had their black spots removed and sectioned for histological purposes. As might be expected with such high dust exposure, the black spots were extremely well-demarcated and followed the lines of lymph flow across and through the parietal pleura.

The most severe and frequently-documented example of pleural response to dust is asbestos pleural plaques. Pleural plaques are commonly seen in asbestos-exposed individuals occurring only on the parietal pleura and diaphragm as discrete, raised, irregularly shaped areas a few millimetres to 10 centimetres in size, having a greyish to ivory white colour depending on their thickness [43]. It is important to note that pleural plaques occur in greatest profusion in exactly the sites where the stomata are in greatest profusion i.e. pleural plaques are '...most commonly found on the posterior wall of the lower half of the pleural spaces, those in the intercostal space tended to have an elliptical shape and ran parallel to the ribs above and below...'. On histological section plaques can be seen to be composed of dense bands or weaves of avascular and largely acellular collagen, with only the occasional fibroblast nucleus to be seen; they are sometimes calcified [44]. These collagenous plaques, whilst commonly seen in association with asbestos exposure are not unique to it, being found following pleural infection or trauma and so can be presumed to be the way that the pleura reacts to injury [44].

Thus there is clear evidence that a proportion of all deposited particles, most commonly urban particulate matter, reach the parietal pleura where they may interstitialise around the stomata and elicit responses. The severity of the response is dependent on the intrinsic toxicity of the dust, with increasing levels of inflammatory/fibrotic response as follows:- soot < mixed mineral dust < short asbestos. The benign nature of asbestos pleural plaque-type responses is evident in the lack of reports of mesothelioma in city dwellers or coalminers despite the prevalence of black spots in these populations and the notable lack of asbestos pleural plaque progression to malignancy [45]. Since normal asbestos pleural plaques are benign and not pre-cancerous, we hypothesise that pleural plaques are a special case of a 'black' spot caused by short asbestos fibres which elicit an unusually florid collagenous response, or as a result of a very high dose of short fibres reaching the peri-stomatal wall. The emphasis on shortness here is important since the key feature of black spots, we contend, is that the particles and short fibres are small enough to negotiate the stomatal openings where they mostly clear to the lymph nodes whilst some interstitialise into the sub-mesothelial interstitium through the proximal lymphatic capillary walls. As described below, this contrasts with events that may occur with long fibres; these cannot negotiate the stomata leading to retention at the stomatal openings, initiating a very different pathobiological sequence of events culminating in a different pathological outcome.

This means that that the primary lesion caused by long fibres must form at the parietal pleura, the site of retention of long fibre dose and therefore the site of response. Mesothelioma would therefore originate not at the visceral pleura but at the parietal pleura. There is good evidence to suggest that this is indeed the case, and numerous studies using thoracoscopy have confirmed that the origin of mesothelioma is the parietal pleura [6]. This is reflected in the staging of mesothelioma which recognises that early mesothelioma is confined to the parietal pleura, while more advanced mesothelioma involves the visceral pleura [48]. Indeed, the prognosis for mesothelioma when it only involves the parietal pleura is much better by around 30 months, than the prognosis arising when mesothelioma involves visceral pleura [6]. This reflects the earliness of the disease stage when it is still confined to the parietal pleura. From a toxicological viewpoint this means that the focus of attention in trying to determine whether any fibre is likely to cause mesothelioma should not be focussed on the question 'Do fibres reach the pleura?' but should be focussed on the question 'Are the fibres retained in the parietal pleura?'.

Fibres found in digested human lung and visceral pleura at autopsy following death from mesothelioma are often short [49] but, as described below, these are not the site to seek the effective fibre 'dose' for mesothelioma, since the parietal pleura is the site of mesothelioma initiation. In fact the site where the effective dose for long fibres is initially applied is the parietal mesothelium, which has seldom been sampled for fibre burden or dimensions. However, in several studies, fibres recovered from the parietal pleura have also been found to be short [50,51]. This may be explained by the fact that, as described above, the location of the longer fibres is likely to be very focal, at the stomata. When this area was specifically sampled in 14 patients diagnosed with asbestos-associated diseases, including mesothelioma, much longer fibres were found concentrated there [5].

In a time course, the long nanotubes caused a sustained high level of inflammation at 7 days, which was the same as was present at 1 day. This is in contrast to the events in the peritoneal space where there is a decline in inflammatory response over 7 days to levels of about one-fifth present on day 1 by day 7. Based on our hypothesis that long fibres were retained at the parietal pleural and that short fibres were not, we used paraffin wax histology to examine the parietal pleural surfaces. In keeping with the hypothesis, long fibres which were visible on day 1 following injection were still visible in granulomas at the surface of the parietal pleura on day 7 (Figure ​(Figure11).11). No short fibres were visible in sections of parietal pleura at 1 or 7 days, however the activated, thickened mesothelium seen at day 1 had returned to normal by day 7 suggesting that the short fibres had been cleared (Murphy, F., Poland, C.A., Ali-Boucetta, H., Al-Jamal K.T., Duffin, R., Nunes, A., Herrero, M-A., Mather, S. J., Bianco, A., Prato, M., Kostarelos, Donaldson, K. Long but not short nanotubes are retained in the pleural space initiating sustained mesothelial inflammation. Submitted for publication).

We therefore hypothesise that the retention of long fires at the stomatal openings on the parietal pleura, coupled with frustrated phagocytosis of pleural leukocytes that attempt to ingest them, produce a chronic pleural mesothelial inflammatory response. Chronic inflammation is known to be a driver for proliferation, genotoxicity, growth factor synthesis and release that are likely to culminate in pathology such as fibrosis, pleural effusion and mesothelioma (Figure ​(Figure1212).