Biopsy for liver cancer: How to balance research needs with evidence‐based clinical practice

Potential conflict of interest: Dr. Sherman and Dr. Bruix consult for Arqule, Bayer, and Celsion. This study has been supported by grants from the Instituto de Salud Carlos III (PI 11/01830). CIBEREHD is funded by Instituto de Salud Carlos III. One of the more active debates in the liver cancer community is whether it is acceptable to base the diagnosis of hepatocellular carcinoma (HCC) on the results of imaging techniques alone or whether biopsy is required. Appropriate clinical use of a biopsy of any organ includes making a diagnosis, determining prognosis (albeit only vaguely), and making a treatment decision that requires knowledge of histology. In the years before availability of noninvasive measures of liver function, liver biopsy was required to determine the extent of fibrosis in hepatitis B and hepatitis C. This was required for assessment of prognosis and for hepatitis C, for predicting response to treatment. In the case of HCC biopsy, is not always necessary for diagnosis. The very high specificity of the typical radiological appearances in the right setting means that the diagnosis can be made without biopsy.1 Furthermore, biomarkers also have a high specificity in the presence of a mass in the liver. Biomarkers are not optimally sensitive, but when present in sufficient concentration can be diagnostic of malignancy (either HCC or intrahepatic cholangiocarcinoma [ICC]).5 This is particularly true in larger tumors and late‐stage disease. The American Association for the Study of Liver Diseases (AASLD) guidelines from 20116 provide an algorithm that has been validated as having high specificity and sensitivity for the diagnosis of HCC.1 Other scientific associations and groups have proposed similar guidelines or have endorsed the AASLD guidelines.6 Biopsy has a place in this algorithm, but only in cases where radiology is not diagnostic. Imaging diagnosis is feasible, but there is a very well‐defined set of requirements: The imaging diagnosis is only feasible in a population at high risk for this malignancy (namely, patients with cirrhosis or long‐lasting liver disease6) because of the high pretest probability that any detected nodule larger than 10 mm in an at‐risk patient is an HCC. The imaging pattern should show intense contrast uptake in the arterial phase, followed by washout in the venous/delayed phase—the images should be acquired on state‐of‐the‐art hardware and read by expert radiologists. With these requirements in place, the pattern described has shown an almost 100% specificity in studies from different centers, with a sensitivity of approximately 60%.1 This means that, in 40% of the patients, the diagnosis will be by biopsy. It is important to stress that even biopsy does not have 100% sensitivity, and when done for small nodules in cirrhosis, the specificity is less good and interobserver agreement is suboptimal.15 These comments serve to frame why diagnostic confirmation before treatment in patients with HCC differs from the general philosophy in oncology. Cancer is unique in medicine in its requirement for histological diagnosis. Histology has, over the years, been important, even essential, for differentiating different cells of origin in tumors arising from an organ (e.g., squamous or adenocarcinoma of the lung) or for differentiating the degree of anaplasia and malignancy of the tumor. Many staging systems require histology to aid treatment decisions. However, the request for a confirmatory biopsy before treatment is usually restricted to first diagnosis. When cancer progresses and recurs after initial surgery, the recognition of metastatic nests in liver, lung, bones, or lymph nodes is usually based on imaging, sometimes accompanied by tumor marker determination (i.e., carcinoembryonic antigen for colorectal cancer). In such a scenario, biopsy is not required because the high probability of malignant involvement resulting from the primary tumor makes the specificity of imaging techniques and tumor markers specific enough for diagnostic purposes. This mimics the high pretest probability scenario already discussed of nodules emerging on the background of a chronic liver disease. This is not the case in the healthy population without liver disease, in whom the risk for HCC is low and therefore pretest likelihood that a lesion is something other than HCC is much higher. The above comments refer to the current recommendations for clinical practice. These should not be seen as being in opposition to research proposals that require tumor biopsy for molecular profiling, given that they are fully complementary. Indeed, translational research will only be fruitful if research investigations are consistent with clinical practice and follow the natural history of the disease. Tissue collections generated through clinical practice in HCC and other tumor types have allowed identification of subsets of tumors that may have druggable targets. Examples are K‐ras in colon,16 anaplastic lymphoma kinase rearrangements in lung,17 and Her2 in breast.18 However, there are no such targets yet identified in HCC. In these examples, the existence of large numbers of resected tumors or of biopsies taken for routine diagnostic purposes allowed investigators to go back and pick out the tumors displaying the mutation and examine how they responded to the specific treatment. After retrospective assessment in tissue collections, the value of such molecular profiling has had to be prospectively validated in research studies that should be expected to have undergone ethics approval. This distinction between clinical practice and research is of paramount importance to frame the current debate about the value of mandatory biopsy in HCC, irrespective of whether the diagnosis was confirmed by robust imaging criteria. Previously, tissue collections used to investigate the correlation of tumor molecular profile with the patient's outcome were originally accumulated during routine clinical management and not for research. Frequently, the use of the material subsequently for research purposes was undertaken without the patient's consent. The extreme example of use of a patient's tissue without consent is the case of Henrietta Lacks (a.k.a Helen Lane), the woman whose tissues resulted in the HeLa cell line, which has been used literally in thousands of research projects, and was essential for the production of for example, the polio vaccine. At the time, this was considered ethical. Subsequently, the Supreme Court in the United States affirmed that discarded tissue belonged to the institution, not to the patient. More recently, greater concern has been expressed that a patient should have the right to refuse that their biological specimen be used for unspecified research purposes. The ethical climate has now changed, and today, if any tissue or body fluid taken for clinical purposes is to be used for research purposes, consent is required, including a specific disclaimer that the patient would not be entitled to any monetary benefit from the research. Further complexity emerges when therapeutic trials, either by independent investigators or pharmaceutical companies, design studies to correlate outcome with tissue profiling or, more important,, when the recruitment of patients is based on an enrichment policy based on the existence of a given characteristic of the tumor cells. Biopsy may be required to perform specific assessments even if the diagnosis had been established years ago. Indeed, given that a tumor's molecular profile may change with disease progression, either spontaneously or because of treatment,19 this may make the information gathered from biopsies at first diagnosis obsolete. Another challenge for those who espouse biopsy is the great degree of heterogeneity within each HCC, particularly larger HCCs.24 There is no guarantee that tissue in the biopsy is representative of the tumor as a whole, complicating evaluation of potential targets for treatment. The suggestion that all HCCs should be biopsied has been defended as necessary for “personalized medicine.” However, at present there are not any established criteria that allow personalized medicine in HCC patients. There are no molecular, metabolic, genetic, or other markers that can be targeted in conventional practice. Novel agents still have to be evaluated within research trials. For example, the presence of the c‐met oncogene in high concentration may indicate a better response to the c‐met inhibitor, tivantinib,25 compared to those tumors expressing low levels of the c‐met oncogene. Phase III studies testing tivantinib (NCT01755767) or cabozantinib (NCT01908426) are ongoing and, if positive, will require that all late‐stage HCCs should undergo biopsy. This is an example of how a biomarker‐driven clinical practice requires prospective protocols (again under ethics control) to validate the suggestions raised at earlier phases of research (others).26 Despite the obvious need to distinguish clinical needs from research needs, there are proponents of liver biopsy for all HCC. The rationale for this is that, by analogy with other cancers (e.g., breast, colon, and melanoma), identification of metabolic pathways that can be modulated for therapeutic effect can possibly be identified by examination of large tissue banks of patients treated with a specific drug, or that common metabolic pathways that might be targets for drugs can also be determined by examining these tissues. As mentioned, no one would argue with the idea that tissue banks are important in the development of new treatments. However, it is clearly unethical outside a research protocol to submit a patient with HCC to a biopsy that is not necessary for clinical purposes. Apart from diagnosis in cases where radiology is nondiagnostic, there are no subtypes or variants of HCC that respond differently to different treatment. Even knowing the state of dedifferentiation is not helpful. Although higher degrees of anaplasia may predict worse outcome, there is, as yet, no evidence to support changing treatment decisions on the basis of histological staging or of any histological marker. If the diagnosis has been established by noninvasive means, performing a biopsy means subjecting the patient to unnecessary risks. There is a death rate associated with biopsy of the liver that, although small, is not zero. In nontumor liver biopsies, the death rate is approximately 1 in 5,000 to 1 in 10,000.28 It is likely to be higher in tumor biopsies, given the vascularity of HCC. Furthermore, although the risk of tumor seeding is low (approximately 2%‐3%),29 this has the potential to transform a curable situation into an incurable one. For researchers, then, the problem is how to accumulate tissue for the kinds of studies that might show that targeted treatment with specific agents may lead to better outcomes in specific subsets of patients. There are two possible pathways. Biopsies that are done because of diagnostic uncertainty can be used. According to the Supreme Court, these are not the property of the patient. However, given that only a minority of patients will need biopsy, these repositories will not be large. They would also potentially be at risk of bias given that the repository would likely include mainly tumors with specific characteristics that make them appear nonspecific on imaging. More practical is to establish a tissue bank. However, this requires the patient's consent, given that the biopsies are being collected specifically for research purposes. Imposing such a policy would have advantages and disadvantages. Obviously, because patients may refuse the biopsy, the number of available specimens will be less numerous. However, a big advantage is that obtaining consent imposes a rigor to the process. Developing a protocol and a consent form and getting it approved by an institutional review board requires some thought about the process so that this is not just a random collection of specimens. Thus, most pharmaceutical company‐sponsored studies today that include storage of tissue or body fluids for future research include, in the consent form, specific consent for that collection and storage. Furthermore, many institutions require that the consent form specify, if not the specific research, at least the type of research that will be done (e.g., look for genetic markers, undertake research related to the drug being tested, and so on). Tissue banks and registries that collect material and data are also required to do so with informed consent. Although there are probably many unregulated data and tissue banks in existence that may have been established without consent, most granting agencies and most research ethics boards will require consent for the collection of these tissues. Finally, one last argument put forward to support routine biopsy is that it is necessary for teaching purposes. This is, at best, controversial. There are literally thousands of HCC specimens already stored in pathology departments all over the world that can be used as teaching material. Also, diagnosis of cases without imaging diagnosis will allow teaching, as will the examination of resected tumors. In summary, the debate around the value of liver biopsy is not a black and white one. Clinical practice needs should not impede research developments by prospective investigations with the needed ethical and informed consent, but research needs should not govern clinical practice. As we have stated in the AASLD guidelines,6 as well as in position papers1 and research manuscripts, the path for further advancement will come from a structured, active blending of clinical practice and research (both in the clinical setting and in the bench). Such productive collaboration is what is in place in referral units and has triggered some of the major advances that we have seen in recent decades.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.