Second Harmonic Generation Reveals Subtle Fibrosis Differences in Adult and Pediatric Nonalcoholic Fatty Liver Disease

Feng Liu, MD; Jing-Min Zhao, MD; Hui-Ying Rao, MD; Wei-Miao Yu, MD; Wei Zhang, MD; Neil D. Theise, MD; Aileen Wee, MBBS, FRCPath, FRCPA; and Lai Wei, MD


Am J Clin Pathol. 2017;148(6):502-512. 

In This Article


SHG/TPEF images enabled in-depth analysis of different collagen features, such as collagen distribution (eg, aggregated vs distributed), morphology (eg, width, length, perimeter, area, and orientation), and location (eg, portal vs central vs perisinusoidal). In this study, we developed qFibrosis quantitative assessment based on SHG/TPEF images, which not only accurately reproduces the NASH CRN stages but also reveals subtle differences and similarities between adult and pediatric collagen deposition.

We first showed the feasibility of qFibrosis quantitative assessment in monitoring liver fibrosis progression in both adult and pediatric NAFLD patients by comparing with the NASH CRN staging system. A q-FP method[22] was reported to measure fibrosis stage in adult patients with NASH by using 70 parameters. In our research, six shared parameters selected from 100 parameters, primarily derived from string collagen, were shown to faithfully recapitulate the fibrosis scores. These string collagen parameters and high AUC values may suggest that these parameters have more discriminative power for reflecting the dynamics of fibrosis progression in adult and pediatric patients with NAFLD, and we may be able to identify the patterns for evaluating fibrosis progression in both groups.

Similarly, qFibrosis quantitative assessment not only reliably reflected liver fibrosis staging but also showed the potential to discriminate finer changes between adult and pediatric patients with NAFLD. Several well-known findings are supported by our data. There is slightly more baseline collagen (ie, normal stroma in both PTs and CVs) in adult livers vs pediatric livers.[26] In terms of disease-associated scarring in NAFLD, we confirm the marked predominance of portal fibrosis in pediatric disease compared with the predominance of central fibrosis in adults. Nonetheless, both populations may display features typical of the other, although to a lesser degree: for example, some pediatric LBx show predominantly central fibrosis, while some adult livers show prominent or predominant portal scarring.

qFibrosis quantitative assessment, however, reveals that similarities of staging using the NASH CRN criteria obscure more subtle differences. For example, while adults may sometimes show portal fibrosis superficially similar to that seen in pediatric LBx, the amount of PT collagen is greater in such adult LBx than is present in similarly staged pediatric LBx. Various possible explanations present themselves. Perhaps the lower baseline stroma of pediatric PTs compared with adult PTs means that the percentage change is more similar. Also, steatohepatitis and steatosis are often more prominent in adult LBx,[27,28] raising the possibility of impedance to sinusoidal blood flow in regions where hepatocytes are enlarged, perhaps, thereby, altering the vascular-stromal interplay in portal regions and in turn leading to increased scar. There is also relatively little information regarding the possible roles played by portal fibroblasts compared with sinusoidal hepatic stellate cells (HSCs) in NAFLD, in either age group;[29–31] there may be differences in activation, numbers, and distribution of such different cell populations that are age dependent.

Another interesting finding is the increased size of CV lumens in pediatric compared with adult NAFLD LBx (supplementary results). This finding suggests the possibility of increased hepatic venous outflow, perhaps reflecting increased portal-central vascular shunting. To our knowledge, this has not been previously investigated. While the possible pathophysiologic implications of such a change are unclear, they may merit investigation; for example, increased arterial flow into central areas in pediatric livers may serve to diminish the likelihood of perivenular and central perisinusoidal scarring. It might also signify a compensatory process contributing to the reported lower incidence of stage 4 (cirrhosis) in pediatric patients compared with adults.[32,33]

Also, in CV regions, there were two patterns of change rather than the single pattern observed for PT. Nearly all aggregated collagen parameters decreased and nearly all distributed collagen parameters increased in pediatric cases from stage 1A/B to 2. This "distributed collagen," as described above, represents the fine collagen fibers of the typical perisinusoidal "chicken wire" pattern of steatofibrosis. Thus, there seems to be a difference in the pathophysiology of this kind of scarring between adults and children. One possible mechanism may include differences between HSC activation in pediatric vs adult livers in NAFLD, but to our knowledge, this also remains an unexplored question. In later stages of disease (stages 3 and 4) in which there are clearly defined central-portal and/or central-central fibrous septa, qFibrosis data confirm a generally increased portal scarring in livers of all ages. This convergence on a common pattern of scarring includes most of the individual qFibrosis parameters. We may consider the possibility that at these stages of disease, the pathophysiology of scarring in NAFLD transitions from disease-specific mechanisms involving the sinusoids, the CVs, and the portal regions to a more generalized process of parenchymal extinction, based on vascular occlusions, as proposed by Wanless and Shiota.[34]

In conclusion, as prior independent studies have shown, pediatric and adult patterns of steatofibrosis differ from each other. In pediatric NAFLD, portal fibrosis predominates, while in adult NAFLD, perisinusoidal fibrosis in centrilobular regions predominates, although these patterns are not exclusive of each other. For such general discriminations, semiquantitative staging such as the NASH CRN system suffices; however, we find that more subtle details of the fine, microscopic distribution and structure of collagen are revealed by SHG. We believe that this difference is analogous to the differences in resolution between, for example, the computed tomography scan and magnetic resonance imaging. In this situation, NASH CRN and similar staging systems are primarily assessed at ×2 to ×4, while SHG assesses changes visible at ×40 to ×100 and therefore captures most of the lower power assessment information but adds abundant additional details. Furthermore, the qFibrosis approach, while replicating the semiquantitative staging of the traditional histologic staging, also provides detailed quantification of features identifiable at both low and high magnification.

NAFLD is a multifactorial disease. The pathogenesis of the condition is incompletely understood. The NASH CRN group reported that two-thirds of cases in children with NAFLD show what that study termed the "adult pattern" of steatosis (zone 3) and one-third the "pediatric pattern" (zone 1).[35] However, in our cohort of 36 pediatric patients, we had only three biopsy specimens showing restricted zone 3 steatosis, and one showed the zone 1 pattern; all others showed panlobular steatosis. Also in our research, 16% incidence for 1A fibrosis, 6% for 1B, and 5% incidence for 1C were observed in adult NAFLD. This incidence was similar to 15% incidence for 1A, 6% for 1B, and 6% incidence for 1C reported by the NASH CRN group.[8] Multiple genetic factors of the host and genetic polymorphisms are recognized as significant factors able to modify the individual risk to develop metabolic and inflammatory processes in NASH. Differences between the histologic aspects of our data and those published from the United States may relate, for example, to differences in ethnicity, genetic factors, environmental exposures, and social behaviors (including socioeconomic status, diet, etc). At least 12 genetic polymorphisms have so far been linked with NAFLD, with reported similarities and differences between Asian and non-Asian populations.[36]

We provide here one example of such analysis—namely, the previously developed qFibrosis assessment, which not only accurately reproduces the NASH CRN staging but also reveals differences and similarities between adult and pediatric collagen deposition not captured by semiquantitative methods. We suggest that this higher resolution may be helpful in the development of hypotheses regarding pathophysiologic mechanisms of scarring in different susceptible populations in NAFLD as well as in other diverse diseases. This higher resolution may also be useful for development and testing of novel therapeutic, antifibrotic interventions. One may expect that different classes of antifibrotic agents may affect different types of collagen or different aspects of collagen production in different sites, by different signaling pathways, and/or by different cell populations. This higher resolution provided by more detailed, quantifiable analysis might thus be useful in preclinical and clinical testing of novel therapeutic agents.