From prediction to judgment: what a five-variable nomogram teaches us about hepatocellular carcinoma

Hepatocellular carcinoma (HCC) remains a disease in which prognosis is determined not by tumour burden alone, but by an uneasy interaction between tumour biology, anatomical extent, hepatic reserve, and treatment feasibility. In this context, Hendi et al. present a timely retrospective cohort analysis of 165 surgically treated patients and propose a practical nomogram for 1-, 3-, and 5-year survival prediction (1). Their final model is built from five readily available variables—body mass index (BMI), albumin, carbohydrate antigen 19-9 (CA19-9), tumour size, and tumour-node-metastasis (TNM) stage and showed strong discrimination at 1, 3, and 5 years (0.838, 0.798, and 0.725, respectively). This is clinically attractive because it favours bedside usability over theoretical complexity. The nomogram is therefore the intellectual centre of the paper: it converts five everyday clinicopathologic signals into an individualized survival estimate rather than leaving clinicians with stage labels alone.

The broader importance of this work is easy to appreciate. Modern HCC management has outgrown a single rigid algorithm. The 2022 Barcelona Clinic for Liver Cancer (BCLC) update reaffirmed the need to integrate tumour burden, liver function, and performance status when recommending therapy (2). Also, clinicians know that two patients with the same stage can travel very different clinical paths. Thus, outcomes predicted by staging systems and experienced by individual patients are not always aligned. Nomograms attempt to narrow the gap between stage and bedside judgment. Their value lies not merely in producing a number, but in helping clinicians discuss risk, choose surveillance intensity, and calibrate therapeutic plans. Used well, such information can also improve conversations with patients and families by converting abstract risk into an individualized estimate that is easier to explain. This can support shared decision-making, align treatment intensity with patient goals, and reduce both unrealistic optimism and therapeutic nihilism. At the same time, a nomogram should be judged not only by discrimination, but by calibration, external validity, and whether it improves real clinical decisions rather than simply producing an elegant number (3).

The most useful way to read the nomogram is not as a mathematical device alone, but as a biologically coherent framework. The five selected variables fall into three domains. First, host reserve is represented by anthropometry (BMI) and serology (albumin). Second, tumour burden is represented by serology (CA19-9). Third, oncological pathology is represented by tumour size and TNM stage. Good HCC prognostic tools should not be tumour-only models, because HCC is not a tumour-only disease, as survival outcomes are also determined by microenvironment and host response to the tumour. A patient with modest tumour burden but frailty or reduced hepatic reserve may do worse than a fitter patient with slightly larger disease. The strength of the nomogram lies less in novelty than in composition: it brings host reserve, tumour burden, and anatomical-pathological extent into a unified clinical framework.

BMI deserves careful interpretation. It would be simplistic to view BMI merely as body habitus. In surgical oncology, low BMI can also be a surrogate for frailty, sarcopenia, chronic inflammation, advanced liver disease, or poor nutritional reserve. Thus, the direction of association is clinically plausible: lower BMI may identify patients who are biologically less able to tolerate disease and treatment. The paper is therefore useful in reminding readers that cancer prognosis is partly determined by the tumour, but partly by the host biological systems. This is also aligned with the growing interest in prehabilitation and physiological optimization before liver surgery (4). In my opinion, both extremes of BMI are related to increased perioperative risk as well as compromised oncological outcomes. High BMI patients are vulnerable to postoperative morbidity, and BMI is reported to have a ‘U’-shaped association with worst outcomes reported in both underweight and obese patients (5).

Albumin is perhaps the most clinically intuitive variable in the nomogram. It is a summary marker of hepatic synthetic function, nutritional state, and systemic illness. In HCC, albumin often carries more prognostic meaning than it first appears to. It tells us something about the liver, something about the host, and something about resilience. That albumin remained in the final model is therefore not surprising. If anything, it confirms that in liver oncology, outcomes are shaped as much by reserve as by resectability. More granular liver reserve tools such as albumin-bilirubin (ALBI) grade have similarly shown prognostic utility across HCC therapies, reinforcing the principle that liver function must remain central to prognostic modelling (6). In a cohort of 285 patients with primary HCC, survival stratified clearly by ALBI grade, with reported 5-year cumulative survival rates of 80% for grade 1, 56% for grade 2, and 23% for grade 3 (7). Notably, even within ALBI grade 2, patients with albumin <3.5 g/dL had significantly worse overall and progression-free survival, more frequent multiple recurrences, and greater serosal invasion, suggesting that albumin may capture tumour aggressiveness as well as hepatic reserve.

CA19-9 is the most thought-provoking variable in the model. CA19-9 should, however, be interpreted cautiously. It is expressed across pancreatobiliary and gastrointestinal epithelia and may be elevated in both malignant and benign inflammatory or cholestatic conditions; accordingly, an abnormal value may reflect tumour biology, biliary obstruction, or both (8). Unlike alpha-fetoprotein (AFP), it is not routinely considered a canonical HCC marker. Its retention in the final nomogram is therefore interesting because it may be capturing biology not fully represented by conventional staging. It suggests that prognostic information may sit in laboratory tests that clinicians often do not expect to fit the typical textbook template. In my opinion, CA19-9 is not specific, and it should be interpreted as a contextual signal rather than a stand-alone decision maker. A normal CA19-9 does not fully reassure, because some individuals are Lewis-antigen non-producers and may not express the marker despite clinically relevant disease (8). This signal is not merely theoretical: in a prospective cohort of 145 patients with HCC, baseline CA19-9 ≥100 U/mL was associated with a 2.7-fold increased risk of death and remained independently significant even after adjustment for Child-Pugh score, AFP, BCLC stage, and model for end-stage liver disease score (9). Interestingly, tissue analysis in that study showed stronger staining in non-tumour liver parenchyma, reactive bile ducts, and progenitor-like cells, raising the possibility that CA19-9 may reflect a permissive diseased liver microenvironment rather than tumour burden alone.

Other low-cost blood-based signals, such as neutrophil-to-lymphocyte ratio and platelet-to-lymphocyte ratio, also merit brief mention because they are inexpensive, easily derived from a routine blood count, and biologically linked to the inflammation-carcinogenesis axis (10,11). Their absence from the present nomogram is not a flaw; rather, it reflects the reasonable discipline of keeping a prognostic tool parsimonious and clinically interpretable. At the same time, these inflammatory indices remain conceptually important because they link prognosis to the systemic inflammation-carcinogenesis axis, which conventional staging alone cannot fully capture.

Tumour size remains important but underemphasized in HCC, particularly because TNM staging tends to treat tumours beyond 2 cm as broadly similar for stage assignment. Size is not everything, but size is rarely nothing. The mathematics alone explains part of its oncologic relevance. Assuming a roughly spherical lesion, tumour volume scales with the cube of diameter (V = π × d³/6): a 1-cm lesion has a volume of only about 0.52 cm³, a 2-cm lesion about 4.19 cm³, and a 3-cm lesion about 14.14 cm³. Increasing diameter from 1 to 3 cm does not merely triple tumour burden; it increases volume and potential cell number by about 27-fold. More cells mean more opportunities for clonal diversification, microenvironmental heterogeneity, hypoxic niches, vascular interface, and treatment-resistant subpopulations, which helps explain why size often reflects biology as well as geometry. In HCC specifically, this is not merely theoretical: microscopic vascular invasion rose from 25% in tumours ≤3 cm to 40% in 3.1–5 cm, 55% in 5.1–6.5 cm, and 63% in tumours >6.5 cm in a large multicentre series (12). Recent meta-analytic data also support the adverse long-term implications of large tumour size after resection, with non-giant HCC showing superior overall and disease-free survival compared with giant HCC (13). The present study therefore adds to the argument that size should not be trivialized within HCC prognostication. It is not merely a diameter measurement; it is often a proxy for tumour age, biological aggressiveness, anatomical distortion, and operative complexity. Importantly, very large tumours also create a different operative problem space: they distort hepatic anatomy, restrict safe hilar dissection and mobilization, increase rupture or seeding risk, and make margin attainment and future liver remnant planning more difficult.

This is reflected in a super-giant HCC series (≥15 cm), where 66.7% had microvascular invasion, 50% developed distant metastasis, and 3-year overall survival was only 29%, underscoring that extreme tumour bulk may signal not only technical difficulty but also biologically adverse disease (14).

TNM stage is the model’s anchor. Among the five variables, it likely carries the greatest structural weight because it summarizes local extent and spreads in a language already familiar to oncologists and surgeons. Yet the interesting point is not that TNM matters—that is expected—but that TNM alone is insufficient. The nomogram becomes more useful precisely because it does not stop at TNM. It adds host condition and biomarker signal. This is an important message for clinicians: anatomy remains foundational, but anatomy alone is not personalized care. Real-world HCC management often diverges from pure stage-based algorithms, as shown in prior Singapore data comparing BCLC and Hong Kong Liver Cancer, where actual treatment frequently differed from recommended pathways (15). The value of the present nomogram is that it bridges the gap between staging frameworks and bedside clinical judgments. This also strengthens the claim for multidisciplinary management of HCC patients to ensure outcomes remain relevant to patients.

Another strength of the paper is that the same five variables also appear to track recurrence-free survival, even though the article title foregrounds overall survival. That coherence is reassuring. That said, recurrence-based endpoints in HCC should still be interpreted with some nuance rather than treated as simple stand-ins for overall survival. Classical work shows that post-resection recurrence is common and biologically heterogeneous (16). Imamura et al. showed that recurrence after hepatectomy is not a single phenomenon: early recurrence was linked more to metastatic-type factors such as microscopic vascular invasion and high AFP, whereas late recurrence tracked more with carcinogenic background factors, supporting the view that recurrence-free survival in HCC blends both relapse and de novo tumorigenesis (17). It suggests that the selected predictors are not random statistical survivors, but clinically meaningful signals that remain relevant across oncologic endpoints. This also resonates with earlier international work showing that large tumour size and liver reserve variables are important in recurrence modelling after resection (18). In practical terms, this means the nomogram may help not only with counselling about lifespan, but also with thinking about follow-up intensity and recurrence vigilance.

The study nonetheless has limitations that should be discussed candidly. It is single-centre, retrospective, and based on a relatively small cohort. The population is heavily weighted toward hepatitis B virus (HBV) infection, cirrhosis, and early TNM stage, with only a small minority having metastasis or multifocal disease. That makes the model useful, but also vulnerable to instability and limited transportability. External validation is necessary before broad adoption. Just as importantly, model evaluation should extend beyond area under the receiver operating characteristic curve alone to include calibration and clinical usefulness, because a nomogram that separates risk groups statistically may still perform poorly when applied to individual patients in different settings. In addition, several clinically important variables were either absent from the final model or not studied, including microvascular invasion, portal hypertension, ALBI-grade style liver reserve metrics, perioperative complications, and post-hepatectomy liver failure (PHLF). Even where remnant volume appears to increase, functional recovery may lag behind morphologic hypertrophy. In an Associating Liver Partition and Portal Vein Ligation for Staged Hepatectomy (ALPPS) case series, standardized future liver remnant increased from 34.4% to 53.0%, yet this did not correlate with indocyanine green retention at 15-minute change, reinforcing the point that volumetry alone may overestimate true functional reserve and therefore underestimate PHLF risk (19). Phosphate kinetics may also have been worth capturing, because post-hepatectomy hypophosphatemia is common, has been variably linked to PHLF, and phosphate replacement has been associated in some cohorts with improved liver function and fewer postoperative complications (20). Likewise, systematic-review data suggest that branched-chain amino acid supplementation reduces postoperative infection, ascites, and length of stay (21), and perioperative steroids reduce overall complications and improve early bilirubin and inflammatory markers (22). This underscores how unreported nutritional and inflammatory modulation may influence postoperative risk without entering the model. These omissions matter because short-term physiology and long-term oncology are closely linked in liver surgery. Prior work has shown that Child-Pugh score, blood loss, PHLF criteria, and early bilirubin rise predict 90-day mortality after hepatic resection for HCC (4). A future-generation nomogram should try to bridge tumour biology and prognosis with liver reserve and treatment response and treatment-related morbidity.

One additional point deserves emphasis. A prognostic model should not be mistaken for a treatment command. Selected HCC patients may still achieve comparable oncologic outcomes with non-resection strategies such as trans-arterial chemoembolization combined with radiofrequency ablation, particularly when morbidity trade-offs are relevant (23). Thus, the present nomogram should be used to sharpen multidisciplinary discussion, not to close it. The proper question is not “What is the score?” but “What does this score mean for this patient’s realistic options?” In that sense, the model is most valuable when used by a multidisciplinary team that understands its clinical context. That distinction is important for safe care. Prediction tools should structure discussion, not substitute for judgment, and their outputs are most useful when interpreted alongside liver reserve, technical feasibility, perioperative risk, and the goals of patients and families.

In summary, Hendi et al. have produced a useful five-variable nomogram that deserves attention not because it is technologically extravagant, but because it is clinically intelligible. The nomogram works precisely because it combines host reserve, tumour signal, and anatomical extent into one readable instrument. BMI and albumin remind us that the patient matters; CA19-9 reminds us that hidden biology matters; tumour size and TNM stage remind us that anatomy still matters. That combination is the central lesson of the paper. In that sense, the present study is part of a wider movement toward personalized HCC care, where the real strength of a nomogram lies in making prognosis more individualized, more discussable, and more clinically actionable at the point of care (24). The next step is not to make prognostic models more complicated for their own sake, but to validate them broadly and enrich them thoughtfully—with liver reserve metrics, recurrence biology, perioperative physiology, and perhaps molecular data—so that prediction can genuinely support judgment rather than merely decorate it.

Acknowledgments

None.

Footnote

Provenance and Peer Review: This article was commissioned by the editorial office, Journal of Gastrointestinal Oncology. The article did not undergo external peer review.

Funding: None.

Conflicts of Interest: The author has completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0286/coif). The author has no conflicts of interest to declare.

Ethical Statement: The author is accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Hendi M, Chen YY, Zhang B, et al. Clinical model for predicting overall survival outcomes in individuals with hepatocellular carcinoma: a retrospective cohort analysis. J Gastrointest Oncol 2026;17:24. [Crossref] [PubMed]
2. Reig M, Forner A, Rimola J, et al. BCLC strategy for prognosis prediction and treatment recommendation: The 2022 update. J Hepatol 2022;76:681-93. [Crossref] [PubMed]
3. Balachandran VP, Gonen M, Smith JJ, et al. Nomograms in oncology: more than meets the eye. Lancet Oncol 2015;16:e173-80. [Crossref] [PubMed]
4. Lei GY, Shen L, Junnarkar SP, et al. Predictors of 90-Day Mortality following Hepatic Resection for Hepatocellular Carcinoma. Visc Med 2021;37:102-9. [Crossref] [PubMed]
5. Berardi G, Kingham TP, Zhang W, et al. Impact of body mass index on perioperative outcomes of laparoscopic major hepatectomies. Surgery 2023;174:259-67. [Crossref] [PubMed]
6. Xu H, Cao D, Zhou D, et al. Baseline Albumin-Bilirubin grade as a predictor of response and outcome of regorafenib therapy in patients with hepatocellular carcinoma: a systematic review and meta-analysis. BMC Cancer 2023;23:1006. [Crossref] [PubMed]
7. Hayata N, Hosui A, Kurahashi T, et al. Impact of Serum Albumin Levels on Prognosis and Recurrence in Patients with Hepatocellular Carcinoma. Cancers (Basel) 2025;17:2971. [Crossref] [PubMed]
8. Lee T, Teng TZJ, Shelat VG. Carbohydrate antigen 19-9 - tumor marker: Past, present, and future. World J Gastrointest Surg 2020;12:468-90. [Crossref] [PubMed]
9. Hsu CC, Goyal A, Iuga A, et al. Elevated CA19-9 Is Associated With Increased Mortality In A Prospective Cohort Of Hepatocellular Carcinoma Patients. Clin Transl Gastroenterol 2015;6:e74. [Crossref] [PubMed]
10. Kabir T, Ye M, Mohd Noor NA, et al. Preoperative Neutrophil-to-Lymphocyte Ratio Plus Platelet-to-Lymphocyte Ratio Predicts the Outcomes after Curative Resection for Hepatocellular Carcinoma. Int J Hepatol 2019;2019:4239463. [Crossref] [PubMed]
11. Shelat VG. Role of inflammatory indices in management of hepatocellular carcinoma-neutrophil to lymphocyte ratio. Ann Transl Med 2020;8:912. [Crossref] [PubMed]
12. Pawlik TM, Delman KA, Vauthey JN, et al. Tumor size predicts vascular invasion and histologic grade: Implications for selection of surgical treatment for hepatocellular carcinoma. Liver Transpl 2005;11:1086-92. [Crossref] [PubMed]
13. Lee AJ, Wu AG, Yew KC, et al. Does size matter for resection of giant versus non-giant hepatocellular carcinoma? A meta-analysis. World J Gastrointest Surg 2023;15:273-86.
14. Wee JJ, Tee CL, Junnarkar SP, et al. Outcomes of surgical resection of super-giant (≥15 cm) hepatocellular carcinoma: Volume does matter, if not the size. J Clin Transl Res 2022;8:209-17.
15. Selby LK, Tay RX, Woon WW, et al. Validity of the Barcelona Clinic Liver Cancer and Hong Kong Liver Cancer staging systems for hepatocellular carcinoma in Singapore. J Hepatobiliary Pancreat Sci 2017;24:143-52. [Crossref] [PubMed]
16. Tung-Ping Poon R, Fan ST, Wong J. Risk factors, prevention, and management of postoperative recurrence after resection of hepatocellular carcinoma. Ann Surg 2000;232:10-24. [Crossref] [PubMed]
17. Imamura H, Matsuyama Y, Tanaka E, et al. Risk factors contributing to early and late phase intrahepatic recurrence of hepatocellular carcinoma after hepatectomy. J Hepatol 2003;38:200-7. [Crossref] [PubMed]
18. Chan AWH, Zhong J, Berhane S, et al. Development of pre and post-operative models to predict early recurrence of hepatocellular carcinoma after surgical resection. J Hepatol 2018;69:1284-93. [Crossref] [PubMed]
19. Chan KS, Shelat VG, Low HM, et al. Is the extent of functional liver remnant increase truly "functional"? A single-institution case series of patients with Associating Liver Partition and Portal vein ligation for Staged hepatectomy (ALPPS). Clin Exp Hepatol 2023;9:28-36.
20. Chan KS, Mohan S, Shelat VG. Outcomes of patients with post-hepatectomy hypophosphatemia: A narrative review. World J Hepatol 2022;14:1550-61. [Crossref] [PubMed]
21. Yap KY, Chi H, Ng S, et al. Effect of perioperative branched chain amino acids supplementation in liver cancer patients undergoing surgical intervention: A systematic review. World J Gastrointest Surg 2023;15:2596-618. [Crossref] [PubMed]
22. Hai HH, Aw P, Teng TZJ, et al. Perioperative steroid administration reduces overall complications in patients undergoing liver resection: A meta-analysis. World J Gastrointest Surg 2021;13:1079-94. [Crossref] [PubMed]
23. Gui CH, Baey S, D'cruz RT, et al. Trans-arterial chemoembolization + radiofrequency ablation versus surgical resection in hepatocellular carcinoma - A meta-analysis. Eur J Surg Oncol 2020;46:763-71. [Crossref] [PubMed]
24. Tan N, Shelat VG. Path towards personalized care of hepatocellular carcinoma: a nomogram for survival prediction. J Gastrointest Oncol 2025;16:1347-50. [Crossref] [PubMed]