Five to seven millimeters of negative tooth-bone discrepancy. A profile that is neither clearly protrusive nor clearly retrusive. Two treatment plans that both look defensible on the models. This is borderline crowding, and it is where extraction mistakes concentrate. Extract, and you risk flattening a face that needed its lip support. Expand, and you risk pushing roots through thin buccal bone into a result that relapses within two years.
Most clinicians resolve the tension by inheriting a philosophy. They become extraction orthodontists or non-extraction orthodontists, and the philosophy makes the decision for them. I see both failure modes in the cases students send me for review, and they fail in opposite directions.
The way out is a decision hierarchy applied in a fixed order: face first, then occlusion, then cephalometrics. Each level answers a different question, which is why the order cannot be shuffled.
Why Rigid Philosophies Fail in Borderline Cases
In severe cases, the decision makes itself. A patient with 12mm of crowding and bimaxillary protrusion will not be treated well without extractions. A patient with 2mm of crowding and a retrusive profile should never lose a premolar.
Borderline crowding is different because the dental discrepancy alone cannot answer the question. The same 6mm of crowding can call for extraction in one face and forbid it in another. A decision process that starts at the occlusion, counting millimeters on the model, leans on the one variable that does not discriminate between the two correct answers.
That is why the face has to come first.
Step 1: The Face Decides the Direction
Before touching a model or a tracing, evaluate the face systematically, in profile and frontal view. Four assessments carry most of the weight in an extraction decision:
- Facial convexity. A convex profile with protrusive dentition tolerates, and often benefits from, the retraction that follows extraction. A straight or concave profile is a warning: retraction here moves the lips in a direction the face cannot afford.
- Nasolabial angle. Already obtuse? Retracting the upper incisors will open it further and age the face. An acute nasolabial angle leaves room for retraction without esthetic cost.
- Lip protrusion and lip support. The lips ride on the incisors. Before you create the space that makes incisor movement possible, ask what each millimeter of that movement will do to lip position.
- Dental exposure. Incisor display at rest and on smiling, checked in both views, tells you how much vertical and sagittal change the esthetics will tolerate.
The output of this step is a direction: either this face can give (retraction is safe, sometimes beneficial), or this face cannot give (the discrepancy must be solved without retracting the profile). Only after that direction is set do the models come out.
Step 2: Quantify the Tooth-Bone Discrepancy
A negative tooth-bone discrepancy can be resolved four ways: extraction, expansion, distalization, or interproximal stripping. In a borderline case the real question is rarely which one. It is how much of each, and where each one stops being safe.
Expansion creates space fast, but bone does not follow the roots indefinitely. Over-expansion pushes roots against the buccal cortical plate, sometimes through it, producing buccal resorption and a result that periodontal biology will not maintain. I treat expansion as a tool with a ceiling, not as an alternative philosophy.
Distalization trades crowding for anchorage demand. It works when the posterior segments have somewhere to go and you have the anchorage to move them there; when neither is true, it fails quietly. Stripping sits at the other end of the scale, resolving small discrepancies at low biological cost. Its role is real but limited: it refines a plan, it does not rescue one.
Extraction resolves large discrepancies definitively. It is also the only one of the four that can simultaneously serve the face, when the facial analysis has already told you retraction is beneficial. By the time you measure anything on the model, Step 1 has quietly weighted all four options for you.
Step 3: Cephalometrics Comes Last
The tracing is last for a reason. Incisor inclinations and skeletal values confirm what the face and the occlusion already suggested, and they warn you about limits — an incisor already retroclined over basal bone cannot be retracted further, no matter how much crowding remains. But a cephalometric number should never override a clear facial finding. Norms describe averages, and the face in front of you may sit far from one.
Executing the Decision: Torque and Anchorage Control
A correct extraction decision can still produce a bad result, because once the space is open, the mechanics decide what the face receives. Rigid philosophies skip this half of the problem, and it is where much of my time in the extractions course goes.
Torque control during retraction
Retraction force applied at the bracket tips crowns lingually and drives roots forward. Uncontrolled, it costs you incisor inclination and deepens the bite, turning a planned bodily retraction into a tipping movement the profile was never planned around. Torque management during space closure is not finishing work; it is the core of extraction mechanics.
Anchorage planning
Every extraction space is shared between anterior retraction and posterior protraction. Decide the ratio before you start, not after the molars have drifted. Where the anchorage demand exceeds what conventional preparation can hold, skeletal anchorage with mini-implants makes the ratio a decision instead of a hope.
Choosing the tooth and sequencing the mechanics
The extraction choice itself should weigh prognosis. A first premolar is not automatically the answer when a second premolar, or a compromised tooth elsewhere, serves the plan better. Be cautious with asymmetric extractions; they shift midlines by design and punish imprecise anchorage.
Then sequence the mechanics in anatomical order: transverse first, then vertical, then sagittal. Retracting into an uncorrected transverse problem builds the error into the result. In borderline cases I also want a digital setup on the table before anything irreversible happens, because watching the planned tooth movement first is worth the time.
This hierarchy, and the mechanics that protect it, is what I teach case by case in Mastering Extractions in Orthodontics, demonstrated on documented clinical cases rather than stated as rules. If you prefer the decision criteria in written form, the Extractions in Orthodontics ebook is the shorter companion.
When the Answer Is Still No Extraction
A face-first protocol is not an extraction-friendly protocol. It rejects extractions as often as it indicates them. A straight profile with an obtuse nasolabial angle and 5mm of crowding pushes you toward combinations of distalization, modest expansion within the bone's limits, and stripping, because the face has nothing to give.
In some borderline cases, the honest output of the analysis is that no camouflage plan serves the patient. The discrepancy is skeletal, and orthognathic surgery is what protects both the face and the stability. Recognizing that early is part of the protocol, not a failure of it.
Your Next Borderline Case
Take the case sitting in your undecided pile and run it in order. Write the four facial findings down rather than glancing at them; I ask students to do this in writing because a quick glance tends to find whatever the preferred philosophy was hoping to see. Let those findings tell you whether retraction would help, hurt, or be neutral for this patient. Then quantify the discrepancy on the models, and only then open the tracing to confirm.
If the face says retraction is beneficial and the discrepancy is real, extraction stops being a philosophical question and becomes a technical plan, starting with which tooth to remove and what anchorage ratio the space closure will follow. If the face says the profile cannot give, the same hierarchy hands you the non-extraction tools with honest limits attached.
Either way, write the reasoning into the chart. When the result is two years old and you are checking stability, that record is what tells you whether the decision or the execution deserves the credit — and it is what sharpens the next call you have to make.