{"id":"27bc9b0a-fc94-48e4-93e2-0c3c6e7f81f8","arxiv_id":"2506.00939","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A multiscale HCV model with infected-cell proliferation and vRNA inheritance reduces to an ODE system and can produce bistability, allowing infection persistence even when R<1.","lead":"This paper builds a mathematical model of hepatitis C infection that tracks viral RNA inside infected liver cells and lets infected cells divide, passing their viral RNA to daughter cells. It shows the model reduces to a simple ODE system, and that infected cell division can keep the infection alive even when the usual reproduction number is below one.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The main ODE reduction (Sec. 3.1) and bifurcation analysis are sound, but the generational-equivalence proof in Sec. 3.2 is invalid as written: Eq (11) drops vRNA transfer terms, so the claim that generation tracking is dynamically irrelevant needs correction.","rationale":"The paper's central result survives the concrete error I found. The Leibniz-rule computation in Sec. 3.1 for R(t) and I(t) is internally consistent, including the crucial cancellation between the vRNA removed from proliferating mother cells and the vRNA inherited by daughters. The resulting ODE system Eq (9) and its logistic specialization Eq (10) are derived correctly under the stated constant-rate assumption, and the Jacobian/eigenvalue analysis of the uninfected and total-infection equilibria checks out. The age-dependence of δ, μ, ρ is a real limitation, but the authors explicitly flag it in the Discussion and the equivalence theorem is framed under constant rates, so it is an acknowledged modeling assumption rather than an internal inconsistency. The genuinely load-bearing defect is in Section 3.2: Eq (11) does not follow from the generational PDE boundary conditions. The R_n equations omit both the gain of vRNA from the previous generation and the loss to the next generation during proliferation. With the corrected transfer terms, the infinite system should still telescope to the same total R equation, so the aggregate conclusion is likely salvageable, but the proof as printed is not valid. Because this error affects a supporting claim about generational dynamics rather than the headline backward-bifurcation result, the appropriate disposition is CONDITIONAL: require the authors to correct Eq (11) and re-verify the equivalence and existence arguments before final acceptance.","tokens_in":26461,"tokens_out":26381,"duration_ms":244423,"concrete_test":"Re-derive Eq (11) from the generational boundary conditions: for n≥1 replace dR_n/dt=αI_n-(δ+μ+ρ)R_n with dR_n/dt=ηb(t)R_{n-1}+αI_n-(δ+μ+ρ)R_n-ηb(t)R_n, and add the missing -ηb(t)R_0 term to the R_0 equation. Then (i) confirm that summing the corrected R_n equations telescopes exactly to the Eq (9) equation for R, and (ii) re-check Proposition A.4's assumptions (linear coefficients a_{i,j} and Dini-uniform convergence) for the corrected system; if the sum does not telescope or the existence proof fails, the generational-equivalence claim must be weakened.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Section 3.2 defines r_n(t,0)=[∫ b h i_{n-1}r_{n-1} da]/[2∫ b h i_{n-1} da] and i_n(t,0)=2∫ b h i_{n-1} da. Differentiating R_n=∫ i_n r_n da along characteristics gives dR_n/dt = i_n(0)r_n(0)+αI_n-(δ+μ+ρ)R_n-∫ b h i_n r_n da. With h=η this is dR_n/dt=ηb R_{n-1}+αI_n-(δ+μ+ρ)R_n-ηb R_n for n≥1, and dR_0/dt=ζβVT+αI_0-(δ+μ+ρ)R_0-ηb R_0. Equation (11) instead states dR_n/dt=αI_n-(δ+μ+ρ)R_n for n≥1 (and omits -ηbR_0 in the n=0 equation), so the inheritable vRNA-transfer terms are absent. The subsequent telescoping argument therefore proves equivalence for a different system in which proliferation neither removes nor transfers vRNA between generations. Since Eq (11) underlies Proposition 3.2 and the claim that generation tracking does not alter dynamics, that proof is invalid as written. The main reduction Eq (4)->Eq (9) in Sec. 3.1 is independent of this argument and appears correct, and the bifurcation analysis of Eq (10) is unaffected; however, the manuscript must be corrected.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper extends the Guedj et al. multiscale PDE model of hepatitis C infection to include proliferation of infected hepatocytes with inheritance of intracellular vRNA. The authors distinguish newly infected cells from proliferation-derived cells, derive a PDE system (Eq. (4)) with boundary conditions encoding vRNA inheritance, and show that, under age-independent intracellular rates and an exponential proliferation hazard, this system is equivalent to the ODE system (Eq. (10)). They then introduce an infinite-generation formulation (Eq. (11)) and argue that tracking generations does not change the total infected-cell and vRNA dynamics. The paper computes the basic reproduction number, analyzes the uninfected, infected, and total-infection equilibria, identifies forward and backward transcritical bifurcations at R=1, and reports a saddle-node transcritical bifurcation, illustrating the results with Matcont and comparing against the model of Elkaranshawy and Ezzat.","tokens_in":26746,"tokens_out":8029,"duration_ms":76970,"significance":"If the technical issues are corrected, this is a useful and broadly sound contribution. The main PDE-to-ODE reduction in Section 3.1 is correct and is derived from first principles, the reproduction number in Eq. (14) is obtained from a characteristic polynomial and matches the local stability condition, and the bifurcation thresholds are given in closed form. The paper also performs a valuable service by demonstrating in Appendix F that a previously published ODE adaptation of the multiscale model spuriously generates intracellular vRNA upon proliferation. The biological message, that a proliferative advantage of infected hepatocytes can create bistability and persistence even when R<1, is clearly supported by the analysis. The main weakness is in Section 3.2, where the derivation of the generation-specific vRNA ODEs is incorrect as written; this affects the claimed proof of equivalence between the generational and non-generational models, although the conclusion appears repairable.","major_comments":[{"comment":"The ODEs for R_n(t) in Eq. (11) omit the vRNA transfer terms that arise from proliferation. Differentiating R_n(t)=∫ i_n(t,a)r_n(t,a)da along characteristics gives, for n≥1, dR_n/dt = ηb(t)R_{n-1} + αI_n - (δ+μ+ρ)R_n - ηb(t)R_n, and for n=0, dR_0/dt = ζβV(t)T(t) + αI_0 - (δ+μ+ρ)R_0 - ηb(t)R_0. The terms -ηbR_n and +ηbR_{n-1} represent the removal of vRNA from generation n when cells divide and its appearance in generation n+1; they do not cancel within a single generation. Equation (11) as printed therefore describes a different system in which proliferation neither transfers nor removes intracellular vRNA between generations. This is a load-bearing error because the subsequent telescoping argument is presented as the proof that generational tracking does not affect dynamics. The conclusion can likely be recovered by including the transfer terms and noting that they telescope when summed over n, provided one also shows R_N(t)→0 as N→∞, but the manuscript must be corrected.","section":"Section 3.2, Eq. (11)"},{"comment":"Proposition 3.2 and the interchange-of-summation argument are stated for the incorrect system Eq. (11), so they do not currently justify the equivalence claim for the generational model derived from the PDEs. In addition, the proof of Lemma A.2 contains an inconsistency: the expression for I_n(t) in Eq. (23) uses the Erlang density with rate 2ηb(t), while Lemma A.2 writes g_n^1(∫_{t-s}^t 2b(σ)dσ) and then uses ∑_{n=1}^∞ g_n^a(s)=a, which changes the prefactor by a factor of η. Once Eq. (11) is corrected, the existence proof and the summation argument need to be redone; it may be simpler to exploit the triangular structure of the corrected I_n subsystem and to bound R_n by a multiple of I_n when proving R_N→0.","section":"Section 3.2 / Appendix A, Proposition 3.2"}],"minor_comments":[{"comment":"The abstract contains the typo 'repoductive number' twice; it should read 'reproductive number'.","section":"Abstract"},{"comment":"The captions for panels C and D both state η=1.26, but the text in Section 4 says the two panels show η=1.26 and η=1.24, respectively; one caption is incorrect.","section":"Figure 2 caption"},{"comment":"The notation for the Erlang density should be harmonized: Eq. (23) and Lemma A.2 use different rate parameters (2ηb versus 2b), and the argument of g_n^1 should be written consistently to make the change of variables transparent.","section":"Appendix A, Eq. (23)"},{"comment":"The sentence 'using the same approach as in Section 3' to compute dR_n/dt hides the boundary term at a=0 and the loss term due to proliferation; it would be helpful to show the intermediate step even in the corrected version.","section":"Section 3.2"},{"comment":"The novelty claim that this is the first saddle-node transcritical bifurcation in a viral dynamics model with only quadratic nonlinearities is strong; the authors should either substantiate it with a more systematic literature search or soften the wording.","section":"Section 4, claim of first SNTC bifurcation"},{"comment":"The text cites Kitagawa et al. as both [28] and [29] for closely related results; the distinction between the PDE-to-ODE paper and the transformed-ODE analysis should be made explicit at the point of citation.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"This is a solid modeling note whose central ODE reduction and bifurcation analysis are sound, and the criticism of the Elkaranshawy-Ezzat model in Appendix F is convincing. The Section 3.2 derivation needs a genuine correction before the equivalence claim can be accepted; the fix is likely local and should not change the main conclusions, but it is not merely typographical. I would be comfortable with publication after the authors address the transfer-term issue and rework the proof of Proposition 3.2."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core of this paper is the multiscale PDE model with infected-cell proliferation and vRNA inheritance, and the ODE reduction in Sec 3.1. That part is sound and useful. The authors correctly show that proliferation conserves total vRNA, so the R equation is unchanged, and they pinpoint why the Elkaranshawy-Ezzat ODE adaptation fails: it invents vRNA at mitosis. The bifurcation analysis of the ODE is also convincing: the backwards transcritical bifurcation at R=1 for eta>1, the bistability with a total-infection equilibrium, and the SNTC point are derived with closed-form conditions and match the numerics. This is a genuine extension of Reluga et al. and, if the SNTC claim holds up, a nice first for a quadratic viral dynamics model.\n\nThe problem is Sec 3.2. Eq (11) is not the correct ODE reduction of the generational PDE system. Differentiating R_n(t)=int i_n r_n da along characteristics gives an extra boundary term i_n(t,0)r_n(t,0)=eta b R_{n-1} plus a loss term -eta b R_n from proliferation of generation-n cells; for n=0 there is an extra -eta b R_0. The printed dR_n/dt = alpha I_n - (delta+mu+rho)R_n drops all of these. The subsequent telescoping argument therefore proves equivalence for a different system in which proliferation neither transfers nor removes vRNA. This is a real error, not a typo, because the whole point of the generational model is to track that transfer.\n\nThat said, the error is fixable and the paper's main conclusion survives. If you add the transfer terms, the sum over generations telescopes exactly: the eta b terms cancel, leaving zeta beta V T + alpha I - (delta+mu+rho)R. So the aggregate dynamics are indeed the same. The proof needs to be redone, with a proper argument that R_N goes to 0 (or that the infinite sums justify the exchange), but the claim is correct.\n\nMinor points: the age-independence assumption is stated up front and acknowledged in the Discussion as a limitation, so I don't hold it against them. alpha=300 is tuned to hit clinical viral loads, but they say so and the qualitative bifurcation structure is robust to parameter changes.\n\nThis paper deserves a serious referee. The main modeling contribution is solid and the error in Sec 3.2 is contained. I'd recommend major revision: fix Eq (11), re-derive the generational equivalence, and re-check the telescoping limit. Who should read it: anyone working on multiscale viral dynamics or proliferation in structured models. I'd take it to our reading group and would cite the Sec 3.1 result if I were writing about HCV modeling.","headline":"Solid multiscale ODE reduction and bifurcation analysis, but the generational-tracking equivalence in Sec 3.2 is invalid as written because Eq (11) drops the vRNA transfer terms; the conclusion still survives correction.","tokens_in":27329,"tokens_out":4620,"would_cite":true,"duration_ms":40160,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["92D30","34C23","35Q92"],"pacs":[],"model":"deepseek-v4-flash","headline":"A multiscale model of hepatitis C shows that infected-cell proliferation can make infection persist even when the basic reproductive number is below one, and reduces exactly to an ODE system whose bifurcation analysis reveals bistability.","keywords":["hepatitis C virus","multiscale viral dynamics","age-structured PDE","infected cell proliferation","viral RNA inheritance","basic reproduction number","backwards bifurcation","saddle-node transcritical bifurcation"],"falsifier":"Simulate the full age-structured model with an age-dependent production or death rate and show that total virus and infected-cell trajectories diverge from the ODE system; any such divergence falsifies the equivalence claim. For the bistability claim, numerically continue the full PDE with $\\eta>1$ and check whether a backward transcritical bifurcation at $R=1$ actually appears.","tokens_in":26226,"feed_emoji":"🦠","tokens_out":10458,"duration_ms":95390,"temperature":0.7,"pith_summary":"This paper shows how to include proliferation of infected hepatocytes in a multiscale model of hepatitis C without breaking the bookkeeping of intracellular viral RNA. The authors construct a structured PDE model in which cells born of division inherit, on average, half the mother cell's vRNA, and prove that this infinite-dimensional model is exactly equivalent to a four-variable ODE system when the intracellular rates are constant. They also prove that an infinite system tracking each generation of divisions collapses to the same ODEs. The bifurcation analysis then shows that if infected hepatocytes divide faster than uninfected ones, the infection can persist even when the basic reproductive number $R$ is below one, because the uninfected equilibrium undergoes a backward transcritical bifurcation and bistability appears.","feed_headline":"Hepatitis C can persist below R=1 when infected cells divide faster","feed_subtitle":"Multiscale model tracking viral RNA inheritance yields bistable infection even below one secondary infection per cell.","key_machinery":"The central mechanism is writing proliferation as a boundary condition rather than a source term. Infected cells are split into newly infected cells, entering at age zero with $\\zeta$ copies of vRNA, and cells born of proliferation, which enter at age zero with half the average vRNA of all proliferating mother cells. Differentiating the total vRNA integral $R(t)=\\int_0^\\infty (r_0 i_0+r_p i_p)\\,da$ along characteristics with Leibniz's rule makes the proliferation terms cancel exactly, leaving the same $R$ equation as in the proliferation-free model. For the generation-tracking version, the infinite system of ODEs telescopes when summed, giving the same aggregate equations. The bifurcation analysis then rests on a quadratic equilibrium condition $F(T^*)=0$ whose two roots can collide only when $\\eta>1$, yielding the backward transcritical bifurcation and, at a codimension-two point, a saddle-node transcritical bifurcation in which a saddle-node and a transcritical bifurcation coincide.","core_discovery":"The paper claims that a multiscale PDE model of hepatitis C that includes proliferation of infected hepatocytes and the inheritance of intracellular vRNA is mathematically equivalent to the ODE system (10) whenever the rates $\\alpha,\\mu,\\rho$ and $\\delta$ are independent of infection age, and that an infinite-generation version collapses to the same equations. Bifurcation analysis of the ODE shows that the uninfected equilibrium $(T_{\\max},0,0,0)$ is stable for $R<1$ in the standard forward case, but when infected hepatocytes have a proliferative advantage ($\\eta>1$) and $R^\\dagger<1$, the transcritical bifurcation at $R=1$ is backward, producing bistability between the uninfected equilibrium and a total-infection equilibrium in which infected cells sustain themselves without new infections. Thus infection can persist even when the basic reproductive number is below one. The paper further identifies a saddle-node transcritical bifurcation that separates the forward and backward regimes and reports that this is the first such bifurcation found in a viral dynamics model with only quadratic nonlinearities.","pith_inferences":["A testable extension suggested by the conservation argument: apply the same boundary-condition inheritance to other cytoplasmic RNA viruses and check whether standard ODE reductions spuriously create intracellular genomes at division.","If the backward bifurcation holds in patients, antiviral therapy that drives $R$ below 1 may still fail to cure when infected hepatocytes have a proliferative advantage; one could experimentally test whether reducing hepatocyte proliferation shrinks the bistable region.","The generational equivalence implies that for population-level kinetics, generation-resolved measurements of intracellular vRNA are not needed; such measurements would be useful only to test the average-inheritance assumption itself.","The saddle-node transcritical bifurcation structure may be generic to within-host models with logistic growth of infected cells, so similar codimension-two points should be looked for in other infection models with quadratic nonlinearities."],"forward_implications":["With $\\eta>1$ and $R^\\dagger<1$, there is a parameter window $R^\\dagger1$ and check whether a backward transcritical bifurcation at $R=1$ actually appears.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the linear-chain transformation of the multiscale HCV PDE into an ODE system that the paper extends to include proliferation."},{"cited_title":"D., Nettles, R","cited_arxiv_id":null,"evidence_quote":"Introduced the standard multiscale model of HCV intracellular vRNA dynamics that this work augments with infected-cell proliferation."},{"cited_title":"C., Dahari, H., and Perelson, A","cited_arxiv_id":null,"evidence_quote":"Found a backward bifurcation in a standard viral dynamics model with infected-cell proliferation, which the paper extends to the multiscale setting."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"A prior ODE adaptation with proliferation that the paper shows spontaneously generates vRNA, motivating the PDE-based derivation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the generational birth-death framework used to justify tracking infected-cell generations."},{"cited_title":"and Wong, R","cited_arxiv_id":null,"evidence_quote":"Supplies the infinite-systems existence result used to prove the generation-tracking model collapses to the aggregate ODEs."},{"cited_title":"and Perelson, A","cited_arxiv_id":null,"evidence_quote":"Gives the stable intracellular vRNA age distribution used for chronic-infection initial conditions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides experimental evidence on proliferation of HCV-infected hepatocytes that motivates the fitness cost or advantage parameter $\\eta$."}],"review_version":1}