Complexity without ethics: Why integration matters for embryo models

This year, the International Society for Stem Cell Research (ISSCR) announced that it would no longer use the terms 'integrated' and 'non-integrated' in its Guidelines for Stem Cell Research and Clinical Translation (see BioNews 1302). This marks a significant shift in the framing of research involving stem cell-based embryo models (SCBEMs), and raises questions about the oversight of this rapidly advancing field.

SCBEMs are structures created in the laboratory from pluripotent stem cells. They are not embryos, but they can mimic important aspects of early human development. Researchers use SCBEMs to study fundamental biological processes such as how cells differentiate, how tissues form, why miscarriages occur, and what might underlie congenital conditions. SCBEMs also hold promise for improving fertility treatments and advancing regenerative medicine.

Because human embryos are scarce and tightly regulated, SCBEMs offer a desirable alternative. Yet their potential brings ethical challenges. At what point should SCBEMs be treated like any other biological material, and at what point should they be subject to similar kinds of scrutiny as embryos?

In pursuit of a sensible answer, until recently the ISSCR Guidelines divided SCBEMs into two groups (see BioNews 1289). 'Non-integrated' SCBEMs reproduced only limited parts of development, such as embryonic tissues without extraembryonic structures. They were thought to lack the potential to progress much further, and so could usually be handled under the same rules as other human tissue.

'Integrated' SCBEMs by contrast, included both embryonic and extraembryonic tissues, and were therefore thought to have at least the hypothetical capacity to mimic the coordinated development of an embryo. Because of this greater potential, they were placed closer to embryos in ethical and regulatory terms, and research involving them required stricter oversight.

This framework had the attraction of clarity, and it offered a measure of overarching disciplinary consensus even if disagreements remained about its finer details. But it was not without critics.

The UK's Code of Practice for the Generation and Use of Human Stem-Cell-Based Embryo Models (the SCBEM Code of Practice) argued that the distinction was unhelpful, pointing out that some models usually considered 'non-integrated' had nonetheless been shown to develop complex, advanced structures (see BioNews 1246). With its latest update, the ISSCR has now taken the same step, retiring the integrated/non-integrated distinction altogether and instead suggesting that oversight be proportionate to the complexity of each model and the specifics of each project (see BioNews 1293).

It is easy to see why the distinction came under pressure. Given that some models described as 'non-integrated' can – despite lacking extraembryonic tissue – develop far further than expected, the label appears misleading. Why not dispense with it altogether?

The answer lies in what integration was meant to capture. Properly understood, the integrated/non-integrated distinction was not just a technical shorthand for the presence or absence of particular tissues. It was a way of tracking whether a model had the biological capacity to reach developmental stages at which morally significant properties – such as sentience, awareness or viability – could emerge. The distinction was always meant as an ethical marker first, with biology providing the evidential basis rather than the ultimate justification.

Therefore, if new evidence shows that SCBEMs once thought to lack morally relevant developmental potential actually possess it, then the sensible response is to reclassify those SCBEMs rather than jettisoning the distinction. When zoologists recognised that whales give live birth and suckle their young, they did not collapse the boundary between fish and mammals – rather, they used this insight to redraw the boundary more precisely.

The alternative suggested by the SCBEM Code of Practice, and now also by the ISSCR, is to place all SCBEM research under case-by-case review and make oversight proportionate to biological complexity. This has an immediate appeal because it seems flexible, tailored, and adaptable to new scientific findings.

Yet on its own, biological complexity is an insufficient standard. Complexity matters only when it is tied to the possible emergence of morally relevant features. If, for example, the shared moral conviction is that higher-order cognitive activity without sentience carries no moral weight – as may be the case with brain organoids (see BioNews 1189, 1223, 1245 and 1300) – then heightened oversight for these models makes little sense, however sublime the complexity may be.

Once complexity is tethered to the possibility of morally relevant features emerging, any oversight committee must ask whether the SCBEM under review has such potential. This is effectively the same as asking whether the SCBEM is integrated or not. No progress has been made yet – the aim of oversight is obscured by the language of biological complexity.

In addition, leaving oversight entirely to case-by-case discretion comes with serious drawbacks. Of course, every project needs ethical evaluation, but those judgments should be anchored in shared normative standards. Such standards should emerge from collective deliberation, where disagreements about moral significance and developmental thresholds are worked into a common framework. If this does not occur, then the same disputes simply resurface in local committees, possibly producing inconsistent outcomes for similar projects.

For instance, one committee might permit research on a gastruloid, judging its complexity irrelevant because it lacks the possibility of sentience, while another committee might block an almost identical project on the grounds that its structural complexity alone warrants stricter scrutiny. In the absence of shared standards, complexity becomes an elastic notion (see BioNews 1253 and 1258) – stretched differently by different reviewers, producing conflicting judgments for similar models. That is not tailored flexibility, it is regulatory arbitrariness.

The ethical robustness of any scientific field waxes and wanes with the presence of shared standards. With standards, oversight shows collective responsibility and clarity about what can and cannot be done. Without standards, the discipline loses the strength of a common voice. The integrated/non-integrated distinction, though imperfect, was an attempt at such a voice. By retiring it, the ISSCR chooses silence over guidance.

These issues may sound abstract but they matter for science, regulation, and society. For researchers, clear guidance provides predictability and a common understanding of what is expected. For oversight bodies, clear guidance enables consistent judgments and reduces the likelihood of conflicting decisions. For the public, the transparency and stability of oversight frameworks are crucial to sustaining trust that SCBEM research is being conducted responsibly.

SCBEMs are advancing rapidly, and regulatory frameworks must evolve in step. The integrated/non-integrated distinction was imperfect, sometimes failing to track scientific realities. Abandoning the distinction altogether, however, creates more problems than it solves. Case-by-case oversight is indispensable, yet on its own it is inadequate. It must be anchored in shared normative standards that clearly define which models cross thresholds of moral concern. Without such standards, oversight becomes fragmented, arbitrary, and less accountable.