Genetics & Precision Medicine Center

Free consultation

Send the report you already hold, and a rough family tree

If any genetic testing has already been done, send the full report as a PDF and not a screenshot of the summary page, because the laboratory name, the panel name and the exact variant notation are all on pages most people never send. If a tumour has been tested or biopsied, send the pathology report and tell us where the paraffin block is stored, how old it is and whether the sample came from bone. Write out the relatives on both sides who have had cancer with the type and the age at diagnosis, since age matters more than number. Tell us where your grandparents came from, and whether your parents were related, because both change how a laboratory should read your result. List current medications and any chemotherapy you have already had.

A report is an opinion with evidence

A blood test for iron gives a number, and the number is the same in every laboratory that measures it properly. A genetic report is a different kind of object. The laboratory reads your sequence, compares it to a standard reference, lists the places where you differ, and then makes an argument about what each difference means. That argument draws on how common the change is in population databases, what computer models predict it does to the protein, whether anyone has tested it in a laboratory, whether it tracks with disease through families, and what has been published. The conclusion is then sorted into one of five boxes.

Those boxes are what the whole service turns on, so they are worth understanding before you are sitting in a room being told which one you are in.

This table scrolls sideways on a narrow screen. Swipe or drag to see every column.

Because the classification is an argument and not a measurement, it can change. Variants get reclassified in both directions as databases grow and as families are studied, and a good laboratory will reissue a report when that happens. Nobody will chase you about it. Asking for reinterpretation every two years, and asking again before any irreversible decision, is something you have to do yourself, and it is the single most useful habit a person with an uncertain result can develop.

Whose genomes the library holds

To decide whether your particular DNA change matters, a laboratory needs to know how often that change turns up in people without the disease. If the change is common in healthy people, it is almost certainly harmless. If it has never been seen before, the laboratory has less to work with and the honest answer is uncertain. So the accuracy of the answer you get depends on whether people who share your ancestry are in the databases at all.

For a large part of the world, they are underrepresented. A study published in 2026 measured how much difference that makes, and it did so in an unusually careful way.

Read that back with a Turkish, Arab, Iranian, Central Asian or North African patient in mind, which describes a great many of the people who come to this hospital, and the practical consequence is clear enough, since the same panel run on the same machine will hand a Middle Eastern patient an uncertain result roughly twice as often as it hands one to a Northern European patient with an identical clinical history. Nothing has gone wrong in the laboratory. The gap is in the library.

What follows from that is a set of working habits. None of it is technology. We record where the four grandparents came from and whether the parents were related, and we put that in the request so the laboratory can weight population frequencies properly. We prefer laboratories that report ancestry-specific frequency data instead of a single pooled figure. We diary uncertain results for reinterpretation instead of leaving them in a drawer. And we say plainly, in the room, that an uncertain result in a patient from an underrepresented population carries less information than the same words would carry in a Swedish patient, which cuts in both directions and is the reason the follow-up matters.

The uncertain result

Here is the rule, and it is not controversial anywhere in clinical genetics. An uncertain variant on its own does not change what happens to you. It does not earn extra scans, it does not earn earlier screening, and it certainly does not earn a preventive mastectomy or a preventive removal of the ovaries. Every serious guideline says the same thing, for the plain reason that a proportion of these variants are later reclassified as harmless, and an operation performed on the strength of one cannot be undone when the letter arrives.

That rule gets broken constantly. It gets broken by patients who read the word variant and hear the word mutation, and it gets broken by clinicians who would rather do something than explain why they are not going to.

Now the complication, because the honest version of this has two halves. Uncertain does not mean harmless. It means unstudied, and how well something has been studied is not evenly distributed across humanity. A study out of a large Californian health system in 2026 makes the point better than any argument could.

Both halves are true at once. Do not act on an uncertain result. Equally, do not file it under nothing to worry about, because the way out of that apparent contradiction is to gather more evidence instead of guessing. Where a family is informative, testing affected and unaffected relatives can show whether the variant travels with the disease. Where the change sits at the edge of a coding region, RNA studies can show whether it disrupts the splicing machinery, which resolves a meaningful share of uncertain calls. And where nothing can be resolved yet, your management is set by your family history and your own clinical picture, which is exactly how it would have been set if the test had never been done.

That last sentence is the one worth taking home.

Inherited risk, and the family

Inherited testing looks at the DNA you were born with, usually from blood or saliva, and it answers a question about your whole life instead of about one tumour. A positive result changes real things. Screening starts earlier and runs more often, and on different equipment. Risk-reducing surgery becomes a conversation with dates in it. Certain drug classes open up, most familiarly the PARP inhibitors in ovarian, breast, pancreatic and prostate cancer. Reproductive options come onto the table for people young enough to want them.

And it stops being only about you, which is the part people underestimate. Every first-degree relative has a one in two chance of carrying the same change, and testing them for that one known variant is quick, simple and definitive in a way that a whole panel never is. That process, working outwards through a family from a known result, is the most cost-effective thing this entire speciality does.

Which leads to a piece of advice that surprises almost everybody who hears it.

Its mirror image is what a clear result actually licenses. Nothing detected on a panel means no known disease-causing change was found in those specific genes. It is not a certificate. If your family history is strong, that history continues to govern your screening no matter how clean the report looks, and any clinician who tells you to relax on the basis of a normal panel in the presence of three affected first-degree relatives has misread the document.

What actionable really means

Tumour profiling is a different test from inherited testing, aimed at a different question, and the two get confused so routinely that it is worth setting them side by side.

This table scrolls sideways on a narrow screen. Swipe or drag to see every column.

Now the number that almost never appears in the advertising for these services, which is the number of people for whom a genomic report ends in a genomic treatment. It is lower than anybody expects.

Almost every tumour has a genetic fault in it. Roughly one patient in ten ends up on a drug chosen because of one. That gap is not a failure of the test, it is a shortage of drugs, and it is the honest frame for the conversation. Set against that, three things a profile does reliably deliver are worth naming. It can rule a drug out, sparing you months of a treatment that was never going to work. It can explain a resistance that appeared out of nowhere. And it can put you in front of a trial that would otherwise never have found you, which is where more than half the matched treatments in that Catalan series came from. Timing matters more than most people are told. A profile ordered when someone is already too unwell to travel to a trial centre or to tolerate a new drug produces an interesting document and nothing else. If profiling is going to be done, the moment to do it is while there is still performance status to spend.

The sample decides the report

Every tumour report arrives looking like a finished document. It rarely says on the front page how much the laboratory was actually able to see, and that is the variable patients are never told about.

Researchers in Japan went back over 2002 profiling tests performed between 2019 and 2025 across thirteen institutions and split them by specimen quality, comparing samples that fully passed the laboratory standard against those that only qualified, and against tests run on blood instead of tissue. Simple single-letter mutations survived poor samples reasonably well. Everything else did not. Detection of copy number changes fell away, and so did the composite signatures that decide immunotherapy eligibility, meaning microsatellite instability, tumour mutational burden and the homologous recombination deficiency signature. In pancreatic cancer, the KRAS mutation that is present in the overwhelming majority of these tumours was found in 93 percent of fully passing samples, 88 percent of merely qualifying ones and 57 percent of blood-based tests. The authors were direct about the consequence, which is that specimen quality changed how many patients were offered a genome-matched therapy at all.

Sit with that pancreatic figure for a moment.

A negative result from a poor sample is not a negative result. It is a sample that could not answer the question, and it is reported in the same typeface as a real answer.

Blood-based testing has genuine strengths and one underappreciated one. A Greek laboratory running a 1021-gene liquid biopsy assay across 1110 patients with metastatic cancer, with contributing centres in Turkey, Romania, Lebanon and the United Kingdom, found an on-label targetable variant in 16.18 percent, rising by a further 40.65 percent once off-label targets, trial-relevant findings and resistance mechanisms were counted. Concordance with tissue testing for on-label targets was 90.34 percent, which is reassuring. The interesting part is what they did alongside. By sequencing the white blood cells in parallel with the plasma, they could tell which variants had come from the tumour and which had come from the patient's own ageing bone marrow, a phenomenon called clonal haematopoiesis that otherwise masquerades as tumour DNA and has sent more than one patient down a pointless treatment path. The same parallel analysis confirmed a genuinely inherited harmful variant in 11.26 percent of the group.

So there are four questions to ask before a tumour test is sent, and none of them are technical.

• How old is the block, and where physically is it now. Blocks more than three to five years old give steadily worse results, and blocks sitting in a hospital in another country take weeks to release.
• How much tumour is actually in it. A biopsy that is mostly normal tissue and scar dilutes the signal below what the assay can read, and a pathologist can tell you this in one look.
• Was the sample taken from bone. Bone specimens are treated with acid to soften them, and that treatment damages DNA badly enough that many laboratories decline the sample outright.
• If blood is being used instead of tissue, are the white cells being sequenced alongside the plasma. Without that, some of what is reported as tumour may not be tumour.

Dosing, and the DPYD problem

Day to day, the most useful genetics has nothing to do with cancer risk and everything to do with dose. Some people carry gene variants that make them clear a drug unusually slowly, so a standard dose behaves like an overdose, and the clearest case is the DPYD gene and the fluoropyrimidine chemotherapies, meaning fluorouracil, capecitabine and their relatives, which are used across bowel, breast, stomach and head and neck cancer. In someone with reduced enzyme activity, a normal dose can produce catastrophic bone marrow failure. Testing before the first cycle is now standard practice in much of Europe, and it should be non-negotiable.

It also carries the same blind spot as everything else on this page.

None of that makes testing pointless. Test. Finding a guideline variant prevents a disaster and costs almost nothing, and the point is simply that a clear panel is no safety certificate, and the first cycle after a clear DPYD result still deserves a close blood count and a low threshold for stopping. Where the ancestry makes the panel a poor fit, measuring enzyme activity directly instead of inferring it from a genotype is a reasonable addition, and it is worth asking for.

The same logic runs through the rest of drug genetics, which is broader than most patients realise. It covers thiopurine dosing before azathioprine or mercaptopurine, irinotecan and the UGT1A1 gene, tamoxifen and CYP2D6, clopidogrel and CYP2C19, and a hearing-loss variant that makes a single dose of a common intravenous antibiotic deafen a child. Each is a one-off test. The answer stays useful for the rest of a life, which is a good argument for having it on file long before the drug is ever needed.

What we do not offer

A genetics service is partly defined by what it refuses, and refusing in public seems more useful than refusing in a consultation room after somebody has bought a flight.

We do not read consumer ancestry kit raw data as though it were a clinical result. People arrive holding a printout from a third-party interpretation website flagging a variant in a serious gene, The anxiety is real and the finding rarely is. When a clinical laboratory went back through 49 samples sent for confirmation of variants found in consumer raw data, 40 percent of those variants turned out to be false positives, and several changes labelled increased risk by the consumer service or by a third-party interpreter were classified as benign by multiple clinical laboratories and were sitting in plain view in population frequency databases as common variation. Those platforms were built to estimate where your ancestors lived. Reading them for disease risk asks them to do something they were never designed for. What we will do is arrange proper confirmatory testing in a clinical diagnostic laboratory, which is the only thing that turns a printout into information.

We do not sell wellness genomics. Panels claiming to optimise your diet, your training programme, your caffeine intake or your skin regime from a cheek swab are not clinical tests, and the evidence supporting the advice they generate is not close to the standard applied to anything else in this hospital.

Nor do we test children for adult-onset conditions where nothing can be done during childhood. A child who carries a BRCA change gains nothing from knowing at nine and loses the ability to decide at twenty-five whether to know at all. Childhood-onset conditions, and conditions where surveillance genuinely begins young, are an entirely different matter and are tested promptly.

And we do not run a test without a conversation before it and a conversation after it. The consent discussion covers what might be found incidentally, who else in the family the answer will implicate, and how the result may affect insurance in whichever country you live in. A laboratory can post a report. Only a service can tell you what to do about it, and that is the entire distinction this page has been making.

What a first appointment involves

More talking than most people expect and less blood-taking. A good deal of it can happen by video before anybody flies anywhere.

Coming to Istanbul

Genetics is the one department in this hospital where the honest advice is often that you need not come at all, or at least not yet.

Genetics travels better than surgery does. The family history interview works perfectly well over video with an interpreter on the call, and Turkish, English, Arabic, Russian and German are covered as standard with other languages arranged in advance. Blood can be drawn in your own city and shipped, provided the courier and the paperwork are organised from this end and never left to you, and paraffin blocks can be requested from your own hospital and forwarded, though customs paperwork for human tissue is the part that most often adds a fortnight, so start it early. Reports come back in English and Turkish, and we translate a clean summary into the language your own oncologist reads.

Where a visit does make sense, two nights covers it and three is comfortable. Most nationalities enter Turkey visa-free or on an electronic visa applied for online in a few minutes, and a letter of invitation from the hospital is issued for anyone whose application needs one. The airport transfer, hotel booking near the hospital and appointments are arranged together so that the consultation, the blood draw and any imaging fall inside a single working week. A companion is welcome throughout and, in this speciality, is more useful than usual, because two people remember a family history better than one person does.

If you are already travelling for surgery here, genetics fits alongside it. Being seen in the same week as your operation planning is efficient, and where a germline result would change the operation itself, which happens in breast and ovarian cancer more than anywhere else, the testing is arranged early enough to actually influence the decision rather than arriving afterwards as a curiosity.

What moves the cost

Genetic testing is quoted in a way that invites comparison of numbers that are not comparable, because the word panel covers everything from three genes to three hundred and the word interpretation covers everything from an automated report to a molecular board.

Scale is the first driver. Testing one known family variant is a small job. A focused panel of the genes relevant to one cancer type is larger. A broad hereditary panel, whole exome sequencing and whole genome sequencing rise from there, and each step upward buys more coverage and more uncertain findings at the same time. Tumour profiling sits in its own bracket again, and a paired test that sequences tumour and blood together, which is how incidental inherited findings are properly separated out, costs more than either alone.

Then come the things that are easy to leave out of a quotation. Whether the counselling appointments before and after are included or billed separately. Whether relatives testing for a known family variant are quoted individually, since a family of six is a different proposition from one person. Whether retrieving, recutting and shipping a paraffin block from another country is inside the figure or outside it, along with customs handling for human tissue. Whether reinterpretation in future years is free or chargeable. Whether a molecular tumour board discussion is part of the service or an extra. Whether the report is translated. Whether a fast-track turnaround, which most laboratories price separately, is what you have been quoted or the standard one. Your own circumstances move it as much as the menu does. A patient with a single known family variant, a clear question and a block sitting in the same building is a straightforward case. A patient with three cancers in the family, no surviving affected relative, a fifteen-year-old block in another country and a previous uncertain result that needs resolving is a different piece of work entirely, and any figure quoted before those facts are known is a guess dressed as a price.

Four questions turn a headline into something comparable. Name the specific test, with its gene count. Then ask whether the figure includes the counselling before and after or only the laboratory work, what it costs to test each relative afterwards, and who pays if the sample fails and has to be repeated, which happens with old blocks frequently enough to be worth settling in advance. All four are answered in writing before anything is booked.

Follow-up once you are home

Genetic results are not an episode of care that finishes. It is a document with a shelf life, and the follow-up is mostly administrative, which is precisely why it gets forgotten.

Three things carry on after you fly home. The first is reinterpretation. If you were given an uncertain result, we write to the laboratory at two years and again at five, and we write immediately if you tell us a new cancer has appeared in the family, because a new case can change the evidence. The second is the family. Cascade testing is easiest while the person who was tested is still engaged and hardest ten years later, so the relative letters are issued at the results appointment and we will speak to relatives directly, in their own language, if that helps them decide. The third is your own screening plan, which needs to be running in your own country rather than in ours. It is written to be usable by a general practitioner anywhere, with intervals, modalities and starting ages stated plainly instead of referred to by guideline number.

For patients whose tumour was profiled, the follow-up is different in character. Trial eligibility shifts as new studies open, and a target that had no drug attached to it two years ago may have one now. If a profile found something interesting that led nowhere at the time, it is worth asking again at each progression rather than treating the original report as closed.

Keep the full report as a PDF somewhere you can find it, alongside the summary letter instead of in place of it, because the exact gene name and the exact variant notation are what any future laboratory or clinician will need, and reconstructing them years later from memory is slow, expensive and sometimes impossible.

Frequently asked questions