Selective fetal growth restriction (sFGR) affects approximately 10%–15% of monochorionic twin pregnancies and is largely attributable to unequal sharing of a single placenta and the presence of unbalanced vascular anastomoses. Early-onset sFGR (onset before 24 weeks of gestation) is relatively uncommon, but clinical management is challenging because of the risks of extreme prematurity. Selective FGR can lead to poor outcomes for both the growth-restricted and normally grown twins, including stillbirth, neonatal death and cerebral palsy.¹ Dilemmas in the management of these pregnancies, which may result in the loss/disability of one or both twins, has a psychological impact on the parents.² In the event of demise of the smaller twin, the risks of death and neurological morbidity to the surviving co-twin are in the order of 15% and 26%, respectively. There are three main clinical management options for early-onset sFGR: expectant management, involving close monitoring but no active intervention; selective termination of the smaller twin, which can enable the larger twin to develop normally; and placental laser photocoagulation to separate the circulations of the twins, thus reducing the risk posed by placental anastomoses to the larger twin.³ Each management option has risks and ethical challenges: expectant management can be psychologically burdensome to the parents, as there is a risk that one or both twins may die; selective termination may not be acceptable to some parents; and placental laser coagulation is not only technically difficult, but of unproven effectiveness in cases of sFGR, unlike other indications like twin-to-twin transfusion syndrome. We have considered ethical precedents from landmark trials in the management of complicated monochorionic twin pregnancies, drawing distinctions with our context where selective termination of the smaller twin adds complex ethical dimensions to patient consent and trial feasibility.^4-6

This situation of significant disagreement and variation in practice within the expert clinical community would seem to establish clinical equipoise, which could best be resolved through a randomised controlled trial to provide evidence-based future practice. However, in cases of sFGR in monochorionic twin pregnancies, the research question becomes more complex because of the heterogeneity in the definition of the condition itself, which may affect the validity of a proposed trial. For example, according to the Delphi consensus, sFGR in monochorionic twin pregnancies is defined as either one solitary parameter (estimated fetal weight [EFW] of one twin <3rd centile) or at least two out of four contributory parameters (EFW of one twin <10th centile, abdominal circumference <10th centile, EFW discordance of ≥25% and umbilical artery Pulsatility Index of smaller twin >95th centile).⁷ This definition varies with professional bodies such as the International Society of Ultrasound in Obstetrics and Gynecology or the American College of Obstetricians and Gynecologists.

Clinical equipoise is usually taken to be a necessary, though not always sufficient, condition for a clinical trial to be ethically justifiable. It is further arguable that if there is no compelling reason not to perform a clinical trial, then clinical equipoise provides a compelling reason why a trial should be done, to resolve uncertainty and improve the effectiveness of treatment for patients in the future, as discussed by Ashcroft et al.⁸ The argument that there is clinical equipoise regarding interventions in early-onset sFGR might be argued to depend on whether we consider that there is one patient (the mother) or two (or in twin pregnancies, more than two). In other words, is an unborn fetus a patient? In jurisdictions where abortion is lawful, the pragmatic answer is that the law considers the pregnant woman to be the only, or the primary, patient.⁹ However, this issue may affect the acceptability of any intervention to pregnant women, even if in general terms there is no ethical objection to selective termination or other interventions in sFGR (Mitchell et al., unpublished observations).¹⁰

The ethical issues relating to clinical trials in pregnancy are complex. Historically, clinical trials in women who even might be pregnant were not permitted, because of potential risks of teratogenesis, miscarriage or stillbirth, or preterm birth. More recently it has been recognised that this approach is unduly restrictive and undermines the development and delivery of effective care to women and their babies, whether they are pregnant or not.^{11, 12} However, the status of trials of treatments specifically aimed at women who are pregnant, or of treatments specifically aimed at the fetus or fetuses, remains to a degree controversial. One reason is that it is not always clear who the patient is (the mother or the fetus or both), and in some countries, it is not clear whether the fetus has moral or legal ‘personhood’, nor whose interests are being protected, as discussed by Chervenak and McCullough.¹³ In the UK, the fetus does not have any legal rights of its own, at least until it is born and has a separate existence from the mother.¹⁴

In practice, ethical guidance focuses on the autonomy of the woman, and her unique position as patient and pregnant woman. Whatever the theoretical basis may be for making judgements about the interests of the future child, it is arguable that only the woman herself is in a position to make those judgements, and it is clear in law and ethics that her informed consent is essential to authorising any clinical trial of a treatment during her pregnancy, regardless of whether the aim of that treatment is to benefit her, her fetus, or both.¹⁴ Although decision-making during pregnancy, especially in situations where the health of the unborn fetus or fetuses is in jeopardy, can be difficult and stressful, we consider that pregnant women are fully autonomous and this must be respected. Hence, in the UK, their consent to participate in any trial remains essential.¹⁵

In the case of a clinical trial of management of early-onset sFGR in monochorionic twin pregnancies, there are further ethical challenges. One is that selective reduction involves termination of pregnancy. The law in the UK does permit termination if one of several conditions is met, one of which is if ‘there is a substantial risk that if the child were born it would suffer from physical or mental abnormalities as to be seriously handicapped’ (Abortion Act 1967 s.1.1.d) [see also s.5(2)].¹⁶ This means that selective termination in a situation of sFGR at an early gestation is lawful in most cases. However, the legality and public acceptability of termination of pregnancy, even on medical grounds such as the health of the pregnant woman or the viability of the fetus, vary considerably between jurisdictions. Therefore, the permissibility of fetal reduction varies, and it may not always be possible to consider it as an intervention in sFGR. In the UK this approach is lawful, and we argue that the state of clinical equipoise relating to the safest, most effective and most acceptable intervention in early-onset sFGR extends to include selective termination of pregnancy in that context. Nevertheless, termination of pregnancy is controversial, and would not be acceptable to all women, even in these circumstances.

In the context of a randomised trial, where selective termination of one fetus is one of the treatment arms, a two-stage consent is required: consent to being randomised, and consent to undergoing selective termination, should that be the treatment arm to which the patient is randomised. Alternatively, a trial design where patients are ‘pre-randomised’, and consent only to the specific treatment they receive and to data collection may be considered.^17-21 However, pre-randomised designs pose the problem that the patient needs to know that they are being randomised without their knowledge or consent, offered participation in a trial, and that other participants in the trial are being offered alternative treatments, before or after randomisation. This takes us to the second ethical issue: patients may have legitimate moral preferences for one treatment or the other, independently of the question of clinical equipoise as to the safety and effectiveness of the different management options. A simple randomisation between options requires both clinical equipoise on the part of the clinical community and acceptance on the part of patients that such clinical equipoise exists, and moral equipoise or indifference as to the moral legitimacy of the management options. This condition of moral equipoise or indifference may be difficult to meet. Again, one solution may be a pre-randomised design, so that the patient has only to consent or refuse consent to the specific option they are offered, and their moral attitude towards it is then part of that consent process. However, this does not solve the problem mentioned above, that they may have a moral preference for a treatment option they are not being offered, but other patients are. Pre-randomisation in trials can help to counter contamination bias. The caveat remains that if most patients opt for pre-randomisation to a particular treatment arm and refuse consent for another, it can be problematic, and the trial may not be feasible. This is a standard issue in pre-randomised designs that may be mitigated by perhaps an ‘intention-to-treat’ analysis.

This may be more challenging in the context of early-onset sFGR, where there may be a higher than anticipated decline rate in allocated treatment options, affecting the intention-to-treat analysis. Some of the potential strategies to enhance patient acceptance of allocated treatments and maintain the integrity of the intention-to-treat principle would include providing comprehensive counselling before consent to ensure informed choices. In pre-randomised designs, although patients are initially allocated to one arm, they maintain the right to withdraw and receive other available standard treatments.

One way out of this conundrum is empirical: assuming that only a minority of patients in this situation have strict and categorical objections to termination of pregnancy, for most patients their decision about what to do will involve some deliberation about the options. The trial design may be ethical in and of itself, but the moral question about whether to accept a termination of pregnancy (or an alternative management strategy that may lead to death or disability of one or both fetuses) can only be answered by each patient for themselves. That being so, there is no simple answer to the question of whether such a trial is ethical or not. It is definitely lawful, but it may or may not be acceptable. This can be determined by empirical research, principally with women who are either going through a monochorionic twin pregnancy complicated by early sFGR (as to whether they would consent to participate in such a trial, and if so, what their reasoning would be), or who have been through such a pregnancy (retrospectively, would they have wanted this option and how would they have chosen if that had been possible). These two categories of women are best placed to help establish how potential trial participants would make decisions about trial participation, and have the clearest understanding of the ethical, moral and psychological issues at stake.

This highlights another issue, which is the psychological burden of choice. In one scenario (expectant management with no other treatment being offered), women may experience the psychological stress of ‘waiting and seeing’, knowing that there is a high probability of an adverse outcome to their pregnancy. In the active treatment scenarios, this psychological stress may be accompanied by the stress of having to make a positive choice and feel the responsibility for the outcome which may go with that. Participation in a trial may also have a psychological benefit, in terms of feeling that one is doing something that may help others in the future, as well as hoping to improve outcomes in their own pregnancy. Nevertheless, such decision-making may also have a psychological burden, in terms of both choosing to participate in a trial and having no direct say in what treatment is then delivered.

Additionally, in some jurisdictions where health care is available only on a paid basis, or other barriers to treatment may exist, there is an ethical risk that some patients may participate in a trial because this is the only way they can access the treatment they prefer. This presents a question of justice. However, in the UK, this is rarely an issue for patients accessing treatment through the National Health Service.

In summary, clinical equipoise between the different management options for early-onset sFGR in monochorionic twin pregnancies and the need for evidence-based clinical practice provide a strong justification for a clinical trial. However, the moral complexities of selective termination and the psychological burden of choice mean that greater than usual weight is placed on the importance of informed consent. Empirical research is needed to explore the moral issues that patients consider when consenting to clinical trials. This research may use the Four Principles approach as the analytical framework and draw on the Declaration of Helsinki for international ethical standards, determining what makes a trial acceptable to patients and feasible to conduct.^{22, 23} Specifically, this aspect of our research is being addressed in Work Package 2 (WP2) of the FERN (Intervention or Expectant Management for Early Onset Selective Fetal Growth Restriction in Monochorionic Twin Pregnancy) study.²⁴ WP2 is dedicated to understanding the factors that influence potential trial participants' willingness to engage in randomised trials, especially in the context of interventions that include ethically challenging options such as selective termination of the smaller fetus.

选择性胎儿生长受限 (sFGR) 影响大约 10%–15% 的单绒毛膜双胎妊娠，很大程度上归因于单个胎盘的不平等共享和血管吻合的不平衡。早发性 sFGR（妊娠 24 周前发病）相对罕见，但由于存在极端早产的风险，临床管理具有挑战性。选择性 FGR 可能会导致生长受限和正常生长的双胞胎出现不良后果，包括死产、新生儿死亡和脑瘫。¹这些怀孕管理中的困境可能会导致一对或一对双胞胎的丧失/残疾，对父母产生心理影响。²如果较小的双胞胎死亡，幸存的双胞胎的死亡和神经发病风险分别约为 15% 和 26%。早发性 sFGR 的临床治疗方案主要有以下三种： 期待治疗，涉及密切监测但不主动干预；选择性终止较小的双胞胎，从而使较大的双胞胎能够正常发育；胎盘激光光凝术分离双胞胎的循环，从而降低胎盘吻合对较大双胞胎造成的风险。³每种处理方案都存在风险和道德挑战：期待处理可能会给父母带来心理负担，因为存在双胞胎中的一个或两个都可能死亡的风险；一些家长可能无法接受选择性终止妊娠；与双胎输血综合征等其他适应症不同，胎盘激光凝固不仅在技术上困难，而且在 sFGR 病例中的有效性尚未得到证实。我们考虑了处理复杂单绒毛膜双胎妊娠的里程碑式试验中的伦理先例，并与我们的背景进行了区分，即选择性终止较小双胞胎的情况为患者同意和试验可行性增加了复杂的伦理维度。^4-6

专家临床界在实践中存在重大分歧和变化的情况似乎会建立临床平衡，最好通过随机对照试验来解决，以提供基于证据的未来实践。然而，对于单绒毛膜双胎妊娠中的 sFGR 病例，由于病情本身定义的异质性，研究问题变得更加复杂，这可能会影响拟议试验的有效性。例如，根据德尔菲共识，单绒毛膜双胎妊娠中的 sFGR 被定义为一个单独参数（一对双胞胎的估计胎儿体重 [EFW] < 3%）或四个贡献参数中的至少两个（一对双胞胎的 EFW <第 10 个百分位，腹围<第 10 个百分位，EFW 不一致≥25%，较小双胞胎的脐动脉搏动指数>第 95 个百分位）。⁷该定义因国际妇产科超声学会或美国妇产科医师学会等专业机构的不同而异。

临床平衡通常被认为是临床试验在伦理上合理的必要条件，但并不总是充分条件。进一步争论的是，如果没有令人信服的理由不进行临床试验，那么临床平衡提供了一个令人信服的理由为什么应该进行试验，以解决不确定性并提高未来患者治疗的有效性，如阿什克罗夫特等人。⁸关于早发 sFGR 的干预措施存在临床平衡的论点可能取决于我们是否认为有一名患者（母亲）或两名患者（或在双胎妊娠中，超过两名患者）。换句话说，未出生的胎儿是病人吗？在堕胎合法的司法管辖区，务实的答案是法律认为孕妇是唯一或主要的患者。⁹然而，这个问题可能会影响对孕妇任何干预措施的可接受性，即使一般而言，对 sFGR 的选择性终止或其他干预措施没有道德上的反对（Mitchell 等人，未发表的观察结果）。¹⁰

与妊娠期临床试验相关的伦理问题很复杂。从历史上看，即使是可能怀孕的女性也不允许进行临床试验，因为存在致畸、流产、死产或早产的潜在风险。最近，人们认识到这种方法具有过度的限制性，并且损害了对妇女及其婴儿（无论是否怀孕）的有效护理的发展和提供。^{11, 12}然而，专门针对怀孕妇女的治疗试验或专门针对胎儿的治疗试验状况在一定程度上仍然存在争议。原因之一是，并不总是清楚患者是谁（母亲或胎儿或两者），并且在一些国家，尚不清楚胎儿是否具有道德或法律“人格”，也不清楚谁的利益受到保护，正如切尔韦纳克和麦卡洛所讨论的那样。¹³在英国，胎儿本身没有任何合法权利，至少在其出生并独立于母亲存在之前是这样。¹⁴

在实践中，道德指导的重点是妇女的自主权以及她作为患者和孕妇的独特地位。无论对未来孩子的利益作出判断的理论基础是什么，可以说只有妇女本人才能做出这些判断，而且法律和道德明确规定，她的知情同意对于授权至关重要。怀孕期间进行的任何治疗临床试验，无论该治疗的目的是使她、她的胎儿还是两者都受益。¹⁴尽管怀孕期间的决策，特别是在未出生胎儿或胎儿的健康受到威胁的情况下，可能会很困难且压力很大，但我们认为孕妇有完全的自主权，这一点必须得到尊重。因此，在英国，他们同意参加任何试验仍然至关重要。¹⁵

在单绒毛膜双胎妊娠早发 sFGR 管理的临床试验中，还存在进一步的伦理挑战。一是选择性减量涉及终止妊娠。英国的法律确实允许在满足几个条件之一的情况下终止妊娠，其中之一是如果“孩子出生后将面临严重的身体或精神异常的风险”（《堕胎法》） 1967 年第 1.1.d 条) [另见第 5(2) 条]。¹⁶这意味着在大多数情况下，在妊娠早期 sFGR 情况下选择性终止妊娠是合法的。然而，即使是出于孕妇健康或胎儿生存能力等医学原因，终止妊娠的合法性和公众可接受性在不同司法管辖区之间也存在很大差异。因此，减胎术的允许性各不相同，并且并不总是可以将其视为 sFGR 的干预措施。在英国，这种方法是合法的，我们认为，与早发 sFGR 最安全、最有效和最可接受的干预措施相关的临床平衡状态延伸到包括在这种情况下选择性终止妊娠。然而，终止妊娠是有争议的，即使在这种情况下，也不是所有妇女都能接受的。

在随机试验中，选择性终止一个胎儿是治疗组之一，需要两阶段同意：同意随机化，并同意接受选择性终止（如果这是治疗组）患者被随机分组​​。或者，可以考虑将患者“预先随机化”并仅同意他们接受的特定治疗和数据收集的试验设计。^17-21然而，预先随机化的设计带来了一个问题，即患者需要知道他们在不知情或不同意的情况下被随机化，被邀请参加试验，并且试验中的其他参与者在之前或之后正在接受替代治疗。随机化后。这将我们带到了第二个伦理问题：患者可能对一种治疗或另一种治疗有合理的道德偏好，与不同治疗方案的安全性和有效性的临床平衡问题无关。选项之间的简单随机化既需要临床界的临床平衡，也需要患者接受这种临床平衡的存在，以及道德平衡或对管理选项的道德合法性漠不关心。这种道德平衡或冷漠的条件可能很难满足。同样，一种解决方案可能是预先随机设计，这样患者只需同意或拒绝同意他们提供的特定选项，并且他们对此的道德态度就是同意过程的一部分。然而，这并不能解决上述问题，即他们可能对他们没有得到的治疗选择有道德偏好，但其他患者却有。试验中的预随机化有助于消除污染偏差。需要注意的是，如果大多数患者选择预先随机分配到特定的治疗组并拒绝同意另一个治疗组，则可能会出现问题，并且试验可能不可行。这是预随机设计中的一个标准问题，可以通过“意向治疗”分析来缓解。

在早发 sFGR 的背景下，这可能更具挑战性，因为分配的治疗方案的下降率可能高于预期，从而影响意向治疗分析。提高患者对分配治疗的接受度并保持意向治疗原则完整性的一些潜在策略包括在同意前提供全面的咨询，以确保知情的选择。在预先随机设计中，尽管患者最初被分配到一只手臂，但他们保留退出和接受其他可用标准治疗的权利。

解决这个难题的一种方法是经验性的：假设在这种情况下只有少数患者严格而明确地反对终止妊娠，对于大多数患者来说，他们关于做什么的决定将涉及对选项的一些深思熟虑。试验设计本身可能是道德的，但关于是否接受终止妊娠（或可能导致一个或两个胎儿死亡或残疾的替代管理策略）的道德问题只能由每位患者回答他们自己。既然如此，这样的试验是否符合伦理的问题就没有简单的答案。这绝对是合法的，但可能会也可能不会被接受。这可以通过实证研究来确定，主要是针对那些正在经历并发早期 sFGR 的单绒毛膜双胎妊娠的女性（关于她们是否同意参加这样的试验，如果愿意，她们的理由是什么），或者曾经经历过这样的怀孕的人（回想起来，他们是否想要这个选择，如果可能的话，他们会如何选择）。这两类女性最适合帮助确定潜在的试验参与者如何做出有关试验参与的决定，并对所涉及的伦理、道德和心理问题有最清晰的了解。

这凸显了另一个问题，那就是选择的心理负担。在一种情况下（不提供其他治疗的预期治疗），女性可能会经历“观望”的心理压力，因为她知道她们的怀孕很可能会出现不良结果。在积极的治疗场景中，这种心理压力可能伴随着必须做出积极选择并对随之而来的结果承担责任的压力。参与试验还可能带来心理上的好处，因为人们会感觉自己正在做的事情可能会在未来帮助他人，并希望改善自己怀孕的结果。然而，这样的决策也可能会带来心理负担，因为既选择参加试验，又无法直接决定随后提供什么治疗。

此外，在某些仅提供付费医疗服务或可能存在其他治疗障碍的司法管辖区，一些患者可能会参与试验，因为这是他们获得自己喜欢的治疗的唯一途径，因此存在道德风险。这就提出了一个正义问题。然而，在英国，对于通过国家卫生服务中心接受治疗的患者来说，这很少是一个问题。

总之，单绒毛膜双胎妊娠早发 sFGR 的不同治疗方案之间的临床平衡以及基于证据的临床实践的需要为临床试验提供了强有力的理由。然而，选择性终止的道德复杂性和选择的心理负担意味着比平常更加重视知情同意的重要性。需要实证研究来探讨患者在同意临床试验时考虑的道德问题。这项研究可以使用四项原则方法作为分析框架，并借鉴赫尔辛基宣言的国际道德标准，确定什么使试验能够被患者接受并可行。^{22, 23}具体而言，我们研究的这一方面正在 FERN（单绒毛膜双胎妊娠早期选择性胎儿生长受限的干预或预期管理）研究的工作包 2 (WP2) 中得到解决。²⁴ WP2 致力于了解影响潜在试验参与者参与随机试验意愿的因素，特别是在包括选择性终止较小胎儿等道德上具有挑战性的选择的干预措施的背景下。