Discussion

Our model demonstrates the population impact of alternative treatment strategies during the landmark era of improved hepatitis C therapies. What is immediately evident from these projections is that no single approach is optimal with regard to all public health-relevant outcomes; in other words, each involves a trade-off. For instance, targeting PWID will have the most advantageous impact on disease transmission, but falls short in terms of averting overt liver disease and death. Conversely, prioritising patients with moderate and advanced liver disease will have the maximum impact on liver-related complications and liver fatalities, but is suboptimal in terms of curtailing incident transmission. In more tangible terms, by doubling treatment uptake and targeting PWID, we have the opportunity to reduce incident infection to negligible levels (i.e. <50 cases a year) in the year 2025. On the flip side, through likewise doubling uptake, but instead prioritising those persons with moderate and advanced fibrosis, we can stabilise SLM (at ∼70 cases per year) by the year 2023.Thus, new therapies afford public health policy makers great opportunities, but equally, pose dilemmas too. Population priorities need to be debated and, ultimately, certain trade-offs accepted.

In an ideal world, the most tolerable and efficacious therapies would be immediately available to every patient in want. Yet, in the face of finite health budgets and ‘sticker-shock’38 drug prices (i.e. the reality today), some system of prioritisation, whereby SVR is pursued with greater urgency in some, is strategically sensible and ethically justifiable. An important finding from this projection model is that liver-related deaths were at their highest when no prioritisation system was in place. In fact, treating 2000 patients randomly (i.e. as per S2000-NP) would have less impact on liver deaths than treating half that number under a fibrosis-prioritising strategy (i.e. as per S1000-ADV); see figure 3. Of course, prioritisation intrinsically means treating less of one group, in order to treat more of another. Here, our strategies are centred on directing therapy away from mildly fibrotic former/never PWID. But how would a policy of shepherding treatment away from this low-priority group work in practice? Particularly so, when there are no drawbacks of toxicity to temper demand (as has always been the case with HCV therapies, until now). Ultimately, the success or failure of any prioritisation approach will rest on whether a societal consensus can be reached as to which patients merit greatest precedence. Yet in parallel, clinicians should be more explicit in articulating to patients what they stand to gain from a SVR—in effect, managing expectations. Until recently, this has been difficult because the prognosis-improvement an SVR confers has been vague (other than perhaps being able to say to patients that SVR lowers the risk of liver complications and death).39 However, a recent simulation model, synthesising the existing literature in a novel and patient-centred way, provides clearer direction. For instance, the 60-year-old patient with mild fibrosis has just a 1.6% and 2.9% probability of gaining additional life years and healthy life years, respectively (where ‘healthy’ is defined as being free from liver decompensation).40 More realistic expectations may lead to patients making more conservative treatment choices if the benefits on offer are accepted to be modest (as has been noted in other areas of medicine41). Measured expectations are needed too, on the part of healthcare policy makers. Our model demonstrates that new therapies, targeted at those with advanced fibrosis, can dramatically reduce and stabilise SLM, nevertheless it will remain, going forward, an important feature of this population. Partly, this reflects a well-noted continued risk of hepatocellular carcinoma among patients with SVR with moderate to advanced fibrosis. More fundamentally however, post-SVR SLM reflects harmful lifestyle exposures intrinsic to the general HCV-infected population (i.e. 90% of HCV infections are acquired through injecting drug use,42 and a history of heavy alcohol use is reported in >1/3 of persons attending treatment services in Scotland43). It is often assumed, that the entirety of liver disease occurring within the chronic HCV-infected population, is de facto attributable to chronic infection per se. In fact, if one examines the health outcomes of persons who spontaneously resolve HCV (i.e. a group that is commensurable to chronic patients in terms of lifestyle exposures, but who crucially, will have harboured viral RNA for a <6-month duration) one finds here too, vastly elevated rates of adverse liver outcomes, relative to population controls.44–46 Our model then, foreshadows the rise of ‘post-SVR’ disease, the corollary of which is that continued engagement with this population to tackle lifestyle factors will be necessary to drive SLM down further, and in effect, win the wider war. To that end, no magic bullet medicines exist; rather an age-old, complex web of psychosocial factors to unpick.

Although the results of this model are intended to be illustrative, we endeavoured to build treatment strategies that were realistic vis-à-vis implementation at a national, or even pan-national, level. However, while increasing the fraction of PWID treated each year by 20% will be seen as achievable, the same scale-up for moderate/advanced fibrotics may be viewed as a greater challenge. PWID, e.g. conveniently attend specific sites; in particular, needle exchanges and opiate substitution centres. Thus, they are readily accessible to outreach initiatives—the implementation of dried-blood spot testing in addiction clinics being one such case in point.47–49 Moderate to advanced fibrotics, on the other hand, are a more diffuse population, that do not similarly gather at any one particular site. As such case-finding is more challenging. To rise to this challenge, we need action at both primary and secondary levels of care. For instance, at the secondary care level, services could prioritise specific individuals known to be at high risk of moderate/advanced fibrosis. From the Scottish clinical database (described previously, e.g. in ref. 22), we can identify >2000 patients with chronic HCV who: (A) have attended a specialist secondary care appointment for management of their infection, (B) are not currently receiving treatment, (C) have an AST Platelet Ratio Index score >0.7 (which is indicative of moderate-advanced cirrhosis50), (D) are alive (as of Nov 2013) and (E) do not have liver decompensation or liver cancer. Targeting these individuals alone would go some way towards fulfilling the treatment quotas set out in S1000-ADV and S2000-ADV. At the primary care level, age-targeted testing in the GP setting51 ,52 is worth consideration (on the logic that patients with advanced fibrosis tend to be older in age). Further, there are many individuals, who although previously diagnosed with HCV, have never gone on to attend a secondary care appointment (in Scotland, we know of 4000 such living persons aged more than 50 years). Could primary care play a greater role in resignposting these specific individuals towards therapy? Overall, our contention is that despite their dispersed nature, we are not powerless to increase treatment uptake in this special interest group. A recent analysis illustrates a steady decline in the fraction of Scottish patients with liver cirrhosis receiving therapy over the last decade.53 The open question then, becomes whether in this decade, and equipped with a bolstered armamentarium, we should aspire to reverse this trend. Alternatively, we may decide that the prospect of curtailing incident infection is a more worthy near-term priority. Clearly, a debate needs to be had.

Our model is parameterised using observational data from the scientific literature and from Scotland's national surveillance databases. It is reassuring, that our modelled outcomes are in line with observational record-linkage data on liver-related sequelae so far (i.e. up until 201314). Yet, there are specific uncertainties and limitations of this model that we must underscore. First, fibrosis progression rates in unselected community patients are underdetermined. Indeed, the overwhelming majority of what we know about the speed of fibrosis accretion emanates from individuals attending specialist services54 who are unlikely to be representative of the wider general-infected population (as per the disease iceberg phenomenon). Second, the real-world efficacy of future therapies is uncertain. In our base case scenario, we assumed treatment is equally effective for cirrhotic and non-cirrhotic genotype 1–2 patients (based on encouraging initial trial data34 ,35). For genotype 3 patients we factored in a 10% differential, yet the emergence of a large efficacy-differential between cirrhotic and non-cirrhotic patients could impugn a fibrosis-prioritising strategy. These uncertainties must be monitored, and models updated accordingly as more robust data becomes emerges. Third, we apply a homogeneous force of incident infection across our PWID population; a considerable simplification of the reality, where the risk of infection will vary according to numerable cofactors. The transmission aspect of our model is indeed more rudimentary than some contemporary models,17 ,55 but is sufficient, we think, to capture the broad trends and trade-offs apparent when choosing a TTP strategy over alternatives. A further complicating scenario that our model does not account for is a rising force of infection over time. This might occur if at-risk individuals perceive HCV to pose a diminished threat. In other words, in the ‘treat and forget’ therapeutic era, PWID may not take the same precautions to avoid infection, if it is regarded as an easily treated condition. This sort of attitude shift may render our projections on declining transmission overly optimistic. Injecting risk behaviours should be closely monitored for signs of change. Fourth, our model is based on the population of Scotland; a typical resource-rich setting where contemporary transmission is driven by injecting drug use. To note, however, that we would not expect our results (i.e. the broad trade-offs outlined here) to extend to settings with different transmission profiles (namely Japan, and low–middle income countries). Fifth, we did not consider the cost-effectiveness of the strategies we examined; rather, only their broad-level impact at the population level in terms of outcomes that are societally important. Although cost-effectiveness data would not trump this work (i.e. there is more to selecting the right treatment approach than simply considering the bottom line cost per QALY), it would add an important dimension, and should be a focus for the future. Sixth, we did not model fibrosis progression according to cofactors, such as older age, male gender and heavy alcohol consumption, but assumed a blanket rate for all. We concede that this could introduce an important bias here because these cofactors accelerate fibrogenesis,56 and may be more prevalent (heavy alcohol use and older age in particular) among those with more advanced disease, than in competing priority groups such as PWID. Thus, in reality, disease progression in these two groups may not occur at the same rate. In this respect, the outlined impact of prioritising moderate/advanced fibrotics on liver-related sequelae might be seen as conservative. Finally, although therapies capable of 80%–95% SVR rates are set to be approved by 2015, ‘sticker shock’ prices38 may mean they are used selectively, with inferior therapies continuing to be prescribed to lower-priority groups (or perhaps to groups for which inferior regimens still offer a reasonable chance of SVR). These eventualities are not considered in our model.

The major strength of this study lies in the combined modelling of incident infection and liver disease progression. By contrast, previous models have tended to consider one or the other, such that they inevitably end up arriving at one-sided view vis-à-vis which patient group is the treatment priority.16 ,17 ,57–60 Combining the two herein is an important advancement that lends crucial insight and a more complete picture for healthcare policy makers. In summary, this work highlights the opportunities but also opportunity costs of new therapeutic advancements. Now is the provident time to carefully consider our population-level priorities and engineer our treatment strategies accordingly.