Apply assisted reproductive technology and ovarian stimulation, including the prevention and management of OHSS

In one line

Assisted reproduction is a controlled sequence — stimulate, retrieve, fertilise, culture, transfer, support, freeze — built around one dominant safety problem, ovarian hyperstimulation syndrome, which modern practice now largely designs out at the front end (antagonist protocol, agonist trigger, freeze-all) rather than rescues at the back end. The consultant task is to deliver a live birth while keeping the woman out of an OHSS bed and out of a high-order multiple pregnancy.

Mechanism & pathophysiology

A natural cycle recruits a cohort of antral follicles each month but, through falling FSH and dominant-follicle feedback, selects a single one to ovulate. Controlled ovarian stimulation overrides that selection: exogenous gonadotrophins (FSH ± LH activity) are given at a dose that keeps the whole antral cohort above the FSH threshold, so multiple follicles grow in parallel. The price of multiplicity is the entire safety problem of the field — more follicles means more granulosa-cell mass, more oestradiol, and after the trigger, more of the vasoactive mediator that drives ovarian hyperstimulation syndrome.

The cycle in sequence. Stimulation runs for roughly 8–12 days under ultrasound and oestradiol monitoring until a lead cohort reaches ~17–18 mm. A trigger then induces the final maturation that nature would have produced from the LH surge — resumption of meiosis, the leading oocytes completing the first meiotic division to metaphase II. Oocytes are aspirated transvaginally under ultrasound guidance about 34–36 hours after the trigger (timed to retrieve mature oocytes before spontaneous ovulation). In the lab, fertilisation is by conventional IVF (motile sperm co-incubated with each oocyte, fertilisation left to the sperm) or ICSI (a single sperm injected directly into the ooplasm, bypassing every natural sperm–oocyte barrier). Embryos are cultured to cleavage stage (day 2–3) or, increasingly, to blastocyst (day 5–6), which both selects the more viable embryos and synchronises better with the endometrium. One embryo is transferred, the luteal phase is pharmacologically supported because stimulation and retrieval disrupt the corpus luteum, and surplus good-quality embryos are vitrified for later frozen transfer.

Why the protocol architecture exists: preventing a premature LH surge. If the woman's own pituitary releases LH before the follicles are mature, she ovulates early and the cycle is lost. Two pituitary-suppression strategies prevent this and define the two protocols. The GnRH-agonist long protocol gives a GnRH agonist from the preceding luteal phase; after an initial flare it downregulates the pituitary, producing deep, reversible hypogonadotrophic suppression before stimulation even begins. The GnRH-antagonist protocol instead starts gonadotrophins on a natural cycle and adds a GnRH antagonist only mid-stimulation (fixed day, or flexible once the lead follicle reaches ~14 mm), giving immediate competitive blockade of the pituitary GnRH receptor. The antagonist protocol is shorter, uses less drug, and — the decisive point — keeps the pituitary responsive, which is what makes the agonist trigger possible.

The trigger is the hinge of the whole safety story. Final maturation can be triggered with hCG, which is structurally LH-like but has a long half-life, so it occupies the corpus luteum's LH receptors for days. In an agonist long-protocol cycle, hCG is the only option, because the pituitary is downregulated and cannot mount its own surge. In an antagonist cycle, you can instead give a GnRH-agonist trigger, which displaces the antagonist and provokes a short, physiological endogenous LH (and FSH) surge from the still-responsive pituitary — final maturation without days of sustained LH-receptor stimulation. That difference is the entire mechanism behind OHSS prevention.

OHSS is a vascular-permeability disease driven by hCG through VEGF. The hyperstimulated, enlarged ovaries — many corpora lutea after retrieval — secrete vascular endothelial growth factor (VEGF) in response to LH-receptor stimulation. VEGF acting on VEGFR-2 increases capillary permeability throughout the body; fluid shifts out of the intravascular space into a "third space" (ascites, pleural effusion, generalised oedema). The intravascular compartment is left contracted: haemoconcentration, a rising haematocrit, oliguria from renal hypoperfusion, electrolyte derangement, and — because the blood is concentrated and the patient often immobile — a markedly raised venous and arterial thromboembolic risk that can persist for weeks. What sustains the whole process is hCG: the LH-receptor occupancy that keeps VEGF flowing comes either from the exogenous hCG trigger (early-onset OHSS, ~3–7 days after trigger) or from endogenous hCG of an implanting pregnancy (late-onset OHSS, ~9–14 days after transfer, typically more severe and prolonged). Remove hCG from the equation and the disease defuses. An agonist trigger removes the exogenous source; a freeze-all (deferring transfer to a later unstimulated cycle) removes the endogenous-pregnancy source. That is why the modern preventive package targets hCG at both ends.

Assessment

The assessment task before ART is to choose the right modality for this couple, to predict ovarian response so the gonadotrophin dose and protocol fit the patient, and to stratify OHSS risk so prevention is built in from the start. This assumes the completed subfertile-couple work-up — ovulation, tubal patency and semen analysis — which is the prior step, not a re-teach here.

Choosing IUI, IVF or ICSI:

• Intrauterine insemination (IUI) ± mild ovarian stimulation suits mild male-factor or unexplained subfertility with at least one patent tube and an adequate post-wash motile count. It is cheaper and less invasive, and in South Africa it is the realistic first-line assisted treatment for many couples for whom IVF is financially out of reach — but per-cycle success is modest and it offers nothing where the tubes are blocked or sperm parameters are poor.
• IVF is indicated for tubal-factor subfertility, endometriosis, unexplained subfertility after failed lesser treatment, and anovulation resistant to ovulation induction.
• ICSI is the answer to a fertilisation problem — significant male factor (severe oligo-astheno-teratozoospermia, surgically retrieved testicular/epididymal sperm), or previous failed/poor fertilisation with conventional IVF. ICSI does not outperform conventional IVF when sperm are normal; using it indiscriminately adds cost and an invasive micromanipulation step for no fertilisation gain. The discipline is to reserve ICSI for the sperm problem it was designed to solve.

Predicting ovarian response — the AMH/AFC axis. Anti-Müllerian hormone (AMH, secreted by small antral and pre-antral follicles) and the ultrasound antral follicle count (AFC) are the two quantitative markers of the recruitable follicle pool, and ESHRE recommends either over older tests (day-3 FSH, inhibin B). They predict the extremes — who will respond poorly and who will respond excessively — far better than they predict an individual live birth, and that is exactly how to use them: to set the gonadotrophin starting dose and to flag the high responder before the first injection. A high AMH and a high AFC, especially with polycystic ovarian morphology, mark the woman who will over-recruit.

OHSS risk stratification is the assessment that changes the whole plan. The high-risk woman is identifiable in advance:

• Patient factors: young age, low body weight, polycystic/metabolic ovarian syndrome (PMOS, previously PCOS), high AMH, high AFC, and a previous OHSS episode.
• Cycle factors (the dynamic warning): a rapidly rising or very high oestradiol, and a large number of growing follicles — a high follicle count on the day of trigger is the single most useful real-time predictor.
• The amplifier: pregnancy. Conception converts a brewing early OHSS into severe, protracted late OHSS through endogenous hCG.

A woman flagged high-risk here is not someone to stimulate cautiously and hope — she is someone whose protocol, trigger and transfer strategy should all be set to the OHSS-avoiding configuration from the outset.

Management

Structure ART management as immediate (the cycle and its prevention) → ongoing (managing established OHSS if it occurs) → long-term (cumulative-conception strategy, fertility preservation and the ethical/access frame). The dominant management decision is not how to treat OHSS but how to not cause it, because the disease is iatrogenic and largely preventable.

Immediate — the cycle and primary OHSS prevention

The preventive package for the high-risk woman is a stacked set of evidence-based choices, each removing a contributor:

• Antagonist protocol over agonist long protocol. It is the safer architecture in the general population and is the enabling choice for the high responder, because it keeps the pituitary responsive to an agonist trigger. ESHRE recommends the antagonist protocol over agonist protocols for comparable efficacy and higher safety.
• Individualised gonadotrophin dosing. Dose to the predicted response (AMH/AFC), not a fixed dose — over-recruitment is the upstream cause of OHSS, so the cheapest prevention is not starting too high.
• GnRH-agonist trigger instead of hCG in the antagonist cycle for any woman at OHSS risk. This is the most powerful single lever: it replaces days of hCG LH-receptor occupancy with a short endogenous surge, near-abolishing early OHSS.
• Freeze-all — vitrify all embryos and defer transfer to a later, unstimulated cycle. This removes the endogenous-hCG (pregnancy) driver of late OHSS and lets the over-stimulated ovaries quiesce. The agonist trigger creates a luteal-phase deficiency that makes a fresh transfer in that cycle perform poorly, so agonist trigger and freeze-all are natural partners.
• Cabergoline — a dopamine agonist (typically 0.5 mg orally daily for ~8 days from the trigger) blunts VEGFR-2 signalling and reduces the incidence and severity of OHSS in at-risk women, a useful adjunct rather than a standalone solution.
• Elective single-embryo transfer (eSET) — discussed below; it is both an OHSS-relevant and a multiple-pregnancy-relevant decision.

The escape valve when stimulation has already over-shot is cycle segmentation: agonist trigger + freeze-all, treating the fresh cycle purely as an oocyte-harvest and deferring all transfer. Cycle cancellation (withholding the trigger entirely) and coasting (withholding gonadotrophins for a day or two while continuing antagonist, letting smaller follicles atrophy and oestradiol fall) are older tools now largely displaced by the agonist-trigger/freeze-all strategy, which preserves the cycle's oocytes rather than wasting them.

Ongoing — managing established OHSS

When OHSS occurs despite prevention, classify severity and treat the physiology — the third-spacing, the haemoconcentration, and above all the thrombosis risk. Severity grading (RCOG Green-top Guideline No. 5) runs mild → moderate → severe → critical:

Management principles:

• Fluids and monitoring. Maintain intravascular volume and urine output — crystalloid, escalating to colloid/human albumin where haemoconcentration and hypoproteinaemia are marked. Monitor haematocrit, U&E/creatinine, albumin, fluid balance and weight. The trap is over-aggressive crystalloid alone, which leaks straight into the third space; the goal is enough volume to perfuse the kidneys without flooding the abdomen.
• Thromboprophylaxis is non-negotiable in moderate–severe disease. Haemoconcentration plus oestrogen plus immobility creates a strongly prothrombotic state, and venous and arterial thromboses (including unusual sites — upper-limb, cerebral) are the events that kill or maim. Prophylactic low-molecular-weight heparin, mechanical prophylaxis, and a low threshold for imaging any focal neurological or limb symptom.
• Paracentesis (transvaginal or transabdominal, ultrasound-guided) drains tense or symptomatic ascites — it relieves respiratory compromise and the abdominal-compartment effect on renal perfusion, and frequently improves urine output and haemoconcentration. Symptomatic tense ascites is drained, not stoically observed.
• The life-threatening complications to anticipate: thromboembolism (the leading cause of death), respiratory compromise from massive ascites/effusion/ARDS, renal failure from sustained hypoperfusion, and ovarian torsion or haemorrhage from the grossly enlarged ovaries. Critical OHSS is an HDU/ICU problem with multidisciplinary input.
• Resolution and the pregnancy fork. If the woman is not pregnant, OHSS is self-limiting and resolves over a week or two as the corpora lutea regress. If she is pregnant, endogenous hCG sustains and worsens it, which is the whole rationale for having frozen all embryos in the high-risk woman — there is then no fresh pregnancy to amplify the syndrome.

Long-term — cumulative strategy, multiples, preservation, third-party reproduction

Freeze-all and eSET reframe success as cumulative, not per-fresh-transfer. Vitrification is good enough that frozen-embryo transfer in a later, more physiological (unstimulated or naturally cycling) endometrium performs at least as well as fresh transfer in many populations — so deferring transfer costs little and buys safety. The corollary is that a single embryo transferred per attempt, with surplus embryos banked, achieves a cumulative live-birth rate across one fresh-equivalent + subsequent frozen transfers that approaches double-embryo transfer without the multiple pregnancy.

The multiple-pregnancy harm is the second great iatrogenic injury of ART, and eSET is its remedy. Twin and higher-order pregnancies after ART carry markedly increased prematurity, growth restriction, pre-eclampsia, perinatal mortality and maternal morbidity — and in a South African setting, where neonatal intensive-care capacity is finite and unevenly distributed, an avoidable ART twin pregnancy can mean a transfer to a distant unit or a baby who does not get a cot. Transferring one good embryo (especially one blastocyst) and freezing the rest is the single most effective way to keep ART success and avoid that harm. The clinical and the ethical arguments point the same way: deliberately accepting a high multiple rate to inflate a per-cycle pregnancy statistic is no longer defensible.

Preimplantation genetic testing (PGT) selects which embryo to transfer:

• PGT-M (monogenic) — for a known single-gene disorder in the family (cystic fibrosis, sickle cell, Huntington, BRCA), testing embryos so an unaffected one is transferred.
• PGT-SR (structural rearrangements) — for a parental balanced translocation/inversion, selecting chromosomally balanced embryos and reducing recurrent loss.
• PGT-A (aneuploidy) — screening otherwise unselected embryos for whole-chromosome aneuploidy to choose a euploid embryo. Its value is contested (developed in the evidence section): it does not reliably raise the live-birth rate per started cycle in a general good-prognosis population, and it adds an embryo biopsy, cost, and the problem of mosaic results. PGT-M and PGT-SR have clear indications; PGT-A does not warrant blanket use.

Fertility preservation (oncofertility) is consultant work that must happen before gonadotoxic treatment. A young woman facing chemotherapy/radiotherapy or fertility-threatening surgery should be offered, where time and disease allow:

• Oocyte or embryo cryopreservation — the established options, requiring a ~2-week stimulation cycle (random-start protocols shorten the wait); embryo freezing needs a partner or donor sperm, oocyte freezing does not.
• Ovarian tissue cryopreservation — laparoscopic harvest and freezing of cortical tissue for later reimplantation; the only option for prepubertal girls and where treatment cannot wait for stimulation, now an established (no longer purely experimental) technique. In oestrogen-sensitive cancers, freeze and avoid a stimulation oestradiol spike (letrozole-supplemented protocols mitigate this).

The SA reality is that oncofertility is concentrated in a few tertiary/private units and is rarely state-funded, so the consultant's duty is to raise it early and refer fast rather than assume access — a missed referral before chemotherapy is an irreversible loss.

Donor gametes and surrogacy widen the options where a woman has no oocytes (premature ovarian insufficiency, Turner syndrome, gonadotoxic treatment), a man has no usable sperm, or a woman cannot carry a pregnancy. In South Africa these are tightly regulated and altruistic only: under the National Health Act 61 of 2003 (Chapter 8) and the Regulations Relating to Artificial Fertilisation, gametes may not be sold (only reasonable compensation), donation is anonymous, and there are statutory limits on use and embryo storage. Surrogacy is governed by Chapter 19 of the Children's Act 38 of 2005 — see the ethics frame below.

Guidelines compared

The major bodies broadly agree on the core safety architecture; the divergences are in emphasis and in funding philosophy.

The recent change worth flagging: NICE NG257 (2026) has superseded the long-standing CG156, and ESHRE's ovarian-stimulation guidance is itself being updated, but the central recommendations — antagonist protocol, agonist trigger and freeze-all for OHSS risk, AMH/AFC for response prediction, single-embryo transfer — are stable and shared. The genuine fault-line is not clinical but economic: international guidelines assume a degree of funded access (cycle entitlements, multiple-cycle pathways) that does not exist for most South African patients, where the binding constraint is cost, not protocol.

The evidence & the controversy

Freeze-all is not a universal good — it is a risk-stratified good. The freeze-all strategy was born from the high responder, and there the case is strongest. In Chen's PCOS trial, electively freezing all embryos and transferring later raised the live-birth rate after the first transfer and cut OHSS roughly five-fold — a clear win in exactly the population most at risk. But generalising "frozen beats fresh" to everyone over-reaches. In ovulatory women without PCOS, Shi found no live-birth advantage to freeze-all, while Wei, in good-prognosis ovulatory women having a single blastocyst, did find a higher singleton live-birth rate with frozen transfer — and Stormlund, in women with regular cycles, found no difference in ongoing pregnancy between freeze-all and fresh. The honest synthesis is that freeze-all reliably prevents OHSS and is the right default for the high responder and the woman on an agonist trigger, but it does not uniformly improve live birth, and it is not free of harm. A consistent and important signal across these trials is a higher rate of pre-eclampsia / hypertensive disease and large-for-gestational-age babies after frozen transfer, plausibly because the frozen-transfer endometrium is prepared in an artificial cycle without a corpus luteum (and its vasoactive products). So the defensible position is selective: freeze-all for OHSS prevention and the agonist-trigger cycle, fresh transfer a reasonable default for the normal responder — not "freeze everything for everyone."

The agonist trigger resolved a real tension. The Cochrane evidence is that a GnRH-agonist trigger reliably reduces OHSS but, in a fresh autologous cycle, lowers live birth — because the short endogenous surge produces a defective luteal phase. For a decade this looked like a trade-off: safety or success. The freeze-all strategy dissolved it. Trigger with an agonist (near-eliminating OHSS), freeze all embryos (removing the failed luteal phase from the equation), and transfer later into a properly prepared endometrium — and Stormlund's freeze-all/agonist-trigger arm was effectively OHSS-free with non-inferior pregnancy outcomes. "Segment the cycle" is now the standard answer for the high responder, and it is the segmentation, not the agonist trigger alone, that makes it work.

Single-embryo transfer is settled where access allows it, and ethically obligatory regardless. The individual-patient-data evidence is unambiguous: elective single-embryo transfer lowers the fresh live-birth rate per transfer but, once an added frozen single transfer is counted, restores the cumulative live-birth rate to near double-embryo-transfer levels — with a multiple-pregnancy rate close to that of spontaneous conception. The only honest argument against eSET is access: where a woman can afford one cycle and one cycle only, the pressure to transfer two embryos to maximise that single chance is real and human. That is a South African pressure specifically, and the consultant's duty is to counsel the cumulative logic and the genuine danger of ART multiples rather than collude with a one-shot mentality — while acknowledging that the funding environment, not the clinical evidence, is what generates the dilemma.

PGT-A is the live controversy. Marketed as a way to "choose the best embryo" and raise success, the randomised evidence undercut the headline: the STAR trial found PGT-A did not improve live birth per intention-to-treat in good-prognosis women, with at best a per-transfer signal in older women that vanished on intention-to-treat analysis. Add the problems of biopsy harm, cost, mosaic results (an embryo labelled "abnormal" that could have made a healthy baby), and the risk of discarding viable embryos, and blanket PGT-A is not evidence-based. It has a defensible niche (recurrent loss, recurrent implantation failure, advanced age with several embryos to choose between) but is over-sold as a routine add-on — a current, commercially-driven controversy the candidate should be able to argue both ways. PGT-M and PGT-SR, by contrast, rest on a clear genetic indication and are not in dispute.

The ethical and access frame is part of the answer, not a footnote. ART in South Africa sits inside a tight legal scaffold and a stark access gradient. Surrogacy under Chapter 19 of the Children's Act 38 of 2005 requires a written agreement confirmed by the High Court before fertilisation, at least one commissioning parent genetically related to the child, permanent and irreversible inability of the commissioning parent to carry a pregnancy, and prohibits commercial surrogacy (reasonable expenses only); a genetically-related surrogate retains a 60-day right to terminate the agreement after birth. Gamete donation under the National Health Act must be altruistic and anonymous. Layered onto this is the affordability problem: IVF is overwhelmingly self-funded, public-sector ART is scarce and waitlisted, and the genuine ethical work is honesty about which evidence-based option the patient can actually access — and not letting a one-cycle budget drive an unsafe two-embryo transfer or a skipped oncofertility referral. A current and contested wrinkle worth handling carefully is the broader anxiety around fertility and environmental endocrine-disrupting exposures: the consultant position is to acknowledge the uncertainty honestly without overstating a causal link that the evidence does not yet support.

Landmark trials & key evidence

A worked number: in Chen's PCOS cohort, OHSS fell from 7.1% to 1.3% with freeze-all — an absolute risk reduction of ~5.8%, so the number needed to treat ≈ 1/0.058 ≈ 17 women managed with a freeze-all strategy to prevent one case of OHSS, alongside a live-birth gain rather than a loss. That favourable arithmetic — safety and efficacy moving together in the high responder — is precisely why freeze-all became the default in PMOS/high-AMH patients but was, correctly, not extrapolated to normal responders where Shi and Stormlund found no live-birth benefit.

Exam traps & red flags

• Treating OHSS as a back-end rescue problem. The modern answer is prevention by design — antagonist protocol, agonist trigger, freeze-all, cabergoline, individualised dosing. Reaching for albumin and paracentesis without first showing you would have prevented it reads as out of date.
• Giving an hCG trigger to the high responder. In a woman flagged high-risk on AMH/AFC/follicle count, hCG is the avoidable error; the agonist trigger (in an antagonist cycle) is the safe choice.
• Forgetting that pregnancy worsens OHSS. Endogenous hCG drives the severe, late, protracted form — which is the whole reason to freeze all embryos in the high responder rather than transfer fresh.
• Omitting thromboprophylaxis in moderate–severe OHSS. Thromboembolism is the killer; haemoconcentration + oestrogen + immobility is a strongly prothrombotic state, and arterial and unusual-site thromboses occur. LMWH is not optional.
• Over-aggressive crystalloid alone. It leaks straight into the third space; volume support escalates to colloid/albumin, and tense ascites is drained.
• Generalising "frozen beats fresh" to everyone. True for OHSS prevention and the high responder; not a universal live-birth upgrade (Shi, Stormlund found no benefit), and frozen transfer carries a higher pre-eclampsia/LGA signal.
• Routine PGT-A as a success-booster. STAR showed no per-cycle live-birth gain in good-prognosis women; biopsy harm, cost and mosaic/discard problems make blanket use unjustified. PGT-M/PGT-SR are different — they have clear genetic indications.
• Using ICSI when sperm are normal. ICSI fixes a fertilisation problem; it adds nothing over IVF with normal sperm and should not be a default.
• Transferring two embryos to maximise a one-shot cycle. The cumulative-success and multiple-pregnancy-harm arguments make eSET the standard; the access pressure is real but is counselled against, not accommodated.
• Missing the oncofertility window. Fertility preservation must be raised and referred before gonadotoxic treatment; a delayed referral is an irreversible loss, and in SA the duty is to refer early because access is scarce.
• Mishandling the SA legal frame. Commercial surrogacy is prohibited and a surrogacy agreement must be confirmed by the High Court before conception under Chapter 19 of the Children's Act; gamete donation is altruistic and anonymous under the National Health Act. Getting the sequence or the commercial prohibition wrong is a medicolegal trap.

Evidence anchors

• Chen et al. — fresh vs frozen embryos in PCOS, N Engl J Med 2016
• Shi et al. — transfer of fresh vs frozen embryos in ovulatory women, N Engl J Med 2018
• Wei et al. — frozen vs fresh single blastocyst transfer in ovulatory women, Lancet 2019
• Stormlund et al. — freeze-all vs fresh blastocyst transfer in women with regular cycles, BMJ 2020
• Youssef et al. — GnRH agonist vs HCG for oocyte triggering in antagonist ART, Cochrane Database Syst Rev 2014
• McLernon et al. — elective single vs double embryo transfer, IPD meta-analysis, BMJ 2010
• Munné et al. (STAR) — PGT-A vs morphology for single frozen-thawed transfer, Fertil Steril 2019
• ESHRE guideline: ovarian stimulation for IVF/ICSI, Hum Reprod Open 2020
• ASRM — prevention of moderate and severe OHSS: a guideline, Fertil Steril 2024
• RCOG Green-top Guideline No. 5 — The Management of Ovarian Hyperstimulation Syndrome (mild/moderate/severe/critical classification; thromboprophylaxis, paracentesis, fluid/colloid resuscitation).
• NICE NG257 — Fertility problems: assessment and treatment (2026, replacing CG156): AMH/AFC ovarian-reserve testing, OHSS-minimising stimulation, single-embryo transfer.
• South Africa — Children's Act 38 of 2005, Chapter 19 (surrogate-motherhood agreements: written, High Court confirmation before fertilisation, genetic link to one commissioning parent, commercial surrogacy prohibited, 60-day termination right for a genetically-related surrogate).
• South Africa — National Health Act 61 of 2003, Chapter 8, and the Regulations Relating to Artificial Fertilisation of Persons (altruistic, anonymous gamete donation; no sale of gametes/embryos; statutory storage limits).