UNIVERSITE DE GENEVE FACULTE DE MEDECINE

Département des Neurosciences Cliniques Professeur P. Montandon et Dermatologie

Clinique et Policlinique d'Oto-rhino- laryngologie et Chirurgie Cervico-Faciale Division de Chirurgie Cervico-Faciale Professeur W. Lehmann

PAROTIDECTOMY COMPLICATIONS. NEW TECHNIQUES FOR THEIR OBJECTIVE EVALUATION, PREVENTION AND TREATMENT

Travail présenté par le

Docteur Pavel Dulguerov

pour obtenir le titre de Privat Docent

Genève

1999 TABLE OF CONTENTS

Table of contents ...... 1 Table of illustrations ...... 4 Acknowledgements ...... 7 SUMMARY ...... 9 1. INTRODUCTION ...... 13 1.1. History of parotid surgery ...... 14 1.2. Anatomy ...... 19 1.2.1. Parotid fascia ...... 24 1.2.2. Contents of the parotid space ...... 29 1.2.3. Histology ...... 30 1.2.4. Parotid lobes ...... 32 1.2.5. Facial nerve anatomy ...... 34 1.2.5.1 Marginal mandibular branches...... 37 1.2.5.2. Cervical branches...... 38 1.2.5.3. Buccal branches...... 38 1.2.5.4. Zygomatic branches...... 38 1.2.5.5. Temporal branches...... 38 1.3. Indications of parotidectomy ...... 41 1.3.1. Histopathology of parotid lesions ...... 42 1.3.2. Preoperative work-up of parotid lesions ...... 44 1.4. Surgical techniques of parotidectomy ...... 45 1.4.1. Parotid surgery mandates ...... 45 1.4.2. Nomenclature of parotid operations ...... 46 1.4.3. General techniques of parotidectomy ...... 48 1.4.3.1. Anesthesia ...... 48 1.4.3.2. Patient positioning ...... 49 1.4.3.3. Infiltration ...... 49 1.4.3.4. Patient preparation ...... 49 1.4.3.5. Draping ...... 49 1.4.3.6. Incision ...... 49 1.4.3.7. The superficial skin flap ...... 50 1.4.3.8. Dissection of the posterior parotid ...... 51 1.4.4. Techniques for facial nerve identification ...... 52 1.4.5. The superficial parotidectomy ...... 57 1.4.6. The total parotidectomy ...... 60 1.4.7. Wound closure ...... 60 2. PAROTIDECTOMY COMPLICATIONS ...... 61 2.1. Facial nerve ...... 62 2.1.1. Historical background ...... 62 2.1.2. Techniques of evaluation of facial nerve function ...... 64 2.1.3. Topographic facial nerve function testing ...... 66 2.1.3.1. Burres' linear measurement studies ...... 66 2.1.3.2. Multi-camera linear measurement studies ...... 71 2.1.3.3. Other linear measurement studies ...... 71 2.1.3.4. Image subtraction techniques ...... 73 2.1.3.5. Miscellaneous techniques ...... 76 2.1.3.6. Conclusions ...... 78

1 2.1.4. Facial nerve paralysis grading classifications ...... 79 2.1.5. Physiopathology of facial nerve paralysis ...... 86 2.1.5.1. Ischemia ...... 86 2.1.5.2. Mechanical trauma – compression injuries ...... 88 2.1.5.3. Mechanical trauma –crushing ...... 89 2.1.5.4. Mechanical trauma –stretching ...... 90 2.1.5.5. Cold injury ...... 91 2.1.5.6. Damage from electrocautery ...... 92 2.1.5.7. Damage from repeated stimulations ...... 93 2.1.5.8. Nerve toxic substances ...... 93 2.1.5.9. Parotidectomy data ...... 93 2.1.6. Incidence of postparotidectomy facial nerve paralysis ...... 94 2.1.7. Prevention of facial paralysis during parotidectomy ...... 99 2.2. Frey syndrome ...... 100 2.2.1. Historical background ...... 100 2.2.2. Etiology ...... 105 2.2.3. Anatomy and physiology ...... 108 2.2.4. Pathogenesis of Frey syndrome ...... 110 2.2.5. Investigation of Frey syndrome ...... 113 2.2.6. Incidence of Frey syndrome ...... 116 2.2.7. Treatment of Frey syndrome ...... 120 2.2.8. Prevention of Frey syndrome ...... 124 2.3. Wound complications ...... 127 2.3.1. Parotidectomy incisions ...... 127 2.3.2. Retromandibular post-parotidectomy depression ...... 130 2.3.3. Salivary fistula and post-parotidectomy wound collections ...... 131 2.3.4. Skin anesthesia ...... 132 2.4. Recurrence ...... 135 2.4.1. Histology and recurrence ...... 135 2.4.2. Initial surgery and recurrence ...... 136 2.4.3. Age at initial surgery and recurrence ...... 138 2.4.4. About small tumors and capsules ...... 138 2.4.5. Malignant degeneration of pleomorphic adenoma ...... 138 2.4.6. Surgery for recurrences and its complications ...... 139 2.4.7. Re-recurrence of pleomorphic adenomas ...... 140 2.4.8. Radiation of pleomorphic adenomas ...... 140 3. OBJECTIVES ...... 141 4. PATIENTS AND METHODS ...... 142 4.1. Population ...... 142 4.2. Surgery ...... 142 4.3. Post-operative evaluation ...... 144 4.4. Videomimicography ...... 146 4.4.1. Description ...... 146 4.4.2. Subjects ...... 149 4.4.3. Normative measures ...... 149 4.4.4. Statistical analysis...... 151 4.5. Facial gustatory sweating evaluation ...... 152 4.5.1. Blotting paper technique ...... 153 4.5.2. Iodine-sublimated paper histogram (ISPH) ...... 154 4.6. Frey syndrome treatment with botulinum toxin ...... 155 5. RESULTS ...... 157

2 5.1. Videomimicography – the best measures ...... 157 5.2. Videomimicography – variability in normals ...... 160 5.2. Videomimicography – correlation with the House-Brackmann scale ...... 161 5.2. Gustatory sweating – normative data ...... 165 5.3. Parotidectomy - General data ...... 167 5.4. Parotidectomy complications - facial nerve paralysis ...... 170 5.4.1. Classification according to the House-Brackmann scale ...... 170 5.4.1.1. Entire population ...... 170 5.4.1.2. Role of sectioning of facial nerve branches ...... 171 5.4.1.3. Patient's age and postoperative facial function ...... 172 5.4.1.4. Type of parotidectomy and postoperative facial function ...... 173 5.4.1.5. Histopathology and postoperative facial function...... 174 5.4.1.6. Size of lesion and postoperative facial function ...... 176 5.4.1.7. Duration of parotidectomy and postoperative facial function ...... 177 5.4.1.8. Type of intraoperative monitoring technique and postoperative facial function ...... 177 5.4.1.9. Patients with poor postoperative facial function ...... 178 5.4.2. Videomimicography results ...... 179 5.5. Parotidectomy complications – Frey syndrome ...... 180 5.6. Parotidectomy – wound complications ...... 183 5.7. Frey syndrome treatment with botulinum toxin ...... 185 5.8. Recurrences ...... 185 6. DISCUSSION ...... 188 6.1. Videomimicography ...... 188 6.1.1. What facial movements? ...... 188 6.1.2. How should the movements be performed ? ...... 189 6.1.3. What should be measured ? ...... 189 6.1.4. How should the facial measurements performed? ...... 190 6.1.4.1. Not impede facial movement and not touching the face...... 191 6.1.4.2. Reproducibility for a given individual, both in normal and pathologic cases. 191 6.1.4.3. Synchronous data from the left and right side of the face...... 192 6.1.4.4. Automatic measurements to avoid manipulation errors and observer bias. ... 192 6.1.4.5. Rapid, simple and low cost...... 192 6.1.4.6. Well tolerated by patients...... 193 6.1.4.7. Absolute values, not just percentages...... 193 6.1.4.8. Stored for later comparison and evaluations...... 193 6.1.4.9. Not require markings on the face...... 193 6.2. Objective evaluation of Frey syndrome ...... 194 6.3. Post-parotidectomy facial nerve paralysis ...... 195 6.4. Frey syndrome: prevention, detection, and treatment ...... 198 6.5. Wound complications ...... 202 6.6. Recurrence ...... 202 7. CONCLUSIONS ...... 204 8. REFERENCES ...... 205 APPENDIX 1 ...... 228

3 TABLE OF ILLUSTRATIONS

Figure number Page

FIGURE 1: SCHEMATIC REPRESENTATION OF THE PAROTID GLAND AS A TRUNCATED QUADRANGULAR PYRAMID...... 19 FIGURE 2: SCHEMATIC REPRESENTATION OF THE PAROTID GLAND...... 22 FIGURE 3: HORIZONTAL SECTION THROUGH THE NECK AT THE LEVEL OF THE PAROTID GLAND...... 23 FIGURE 4: SCHEMATIC REPRESENTATION OF THE FASCIAL LAYERS OF THE NECK ...... 26 FIGURE 5: SCHEMATIC REPRESENTATION OF THE FASCIAL LAYERS AT THE LEVEL OF THE HARD PALATE ...... 27 FIGURE 6: ANATOMICAL CROSS-SECTION SHOWING THE FASCIAL LAYERS LATERAL TO THE PAROTID GLAND ...... 28 FIGURE 7: SCHEMATIC REPRESENTATION OF THE VESSELS AND NERVES WITHIN THE PAROTID SPACE...... 31 FIGURE 8: SCHEMA OF THE DEVELOPMENT OF THE FACIAL NERVE AND PAROTID GLAND ...... 33 TABLE I: SCHEMATIC DESCRIPTION OF THE BRANCHING PATTERNS OF THE FACIAL NERVE ...... 36 FIGURE 9: BRANCHING PATTERNS OF THE EXTRATEMPORAL FACIAL NERVE ...... 40 TABLE II: DISTRIBUTION OF SALIVARY GLAND TUMORS (BENIGN AND MALIGNANT) IN THE THREE MAIN SALIVARY GLANDS...... 41 TABLE III: INCIDENCE OF MALIGNANCY IN SALIVARY GLAND TUMORS IN THE THREE MAIN SALIVARY GLANDS...... 42 TABLE IV: PATHOLOGICAL CLASSIFICATION OF PAROTID TUMORS, THEIR INCIDENCE (IN PERCENT) AND BRIEF DESCRIPTION OF THEIR TREATMENT ...... 43 FIGURE 10: POSITIONING OF THE PATIENT FOR PAROTID SURGERY ...... 55 FIGURE 11: TYPICAL PAROTIDECTOMY INCISION ...... 55 FIGURE 12: DISSECTION OF THE POSTERIOR PORTION OF THE PAROTID GLAND ...... 56 FIGURE 13: EXPOSURE AND LANDMARKS FOR IDENTIFICATION OF THE FACIAL NERVE TRUNK ...... 56 FIGURE 14: DISSECTION OF FACIAL NERVE BRANCHES DURING SUPERFICIAL PAROTIDECTOMY ...... 59 FIGURE 15: THE OPERATIVE FIELD AT THE END OF SUPERFICIAL PAROTIDECTOMY ...... 59 TABLE V: CLASSIFICATION OF FACIAL EVALUATION SYSTEMS AND THEIR USEFULNESS ACCORDING TO THE DEGREE OF RESIDUAL FACIAL MOTOR FUNCTION ...... 65 FIGURE 16: FACIAL LANDMARKS FOR LINEAR MEASURES ...... 69 TABLE VI: FACIAL MOVEMENTS AND THE MOST MEANINGFUL MEASURES TAKEN (LANDMARKS) ...... 70 TABLE VII: VARIABILITY OF MOUTH AND EYE MOVEMENTS IN 11 NORMAL SUBJECTS USING MICROSCALING ...... 77 TABLE VIII: BOTMAN AND JONGKEES – A GROSS FACIAL NERVE PARALYSIS CLASSIFICATION...... 80 TABLE IX: PEITERSEN – A GROSS FACIAL NERVE PARALYSIS GRADING CLASSIFICATION...... 81 TABLE X: ADOUR AND SWANSON – A REGIONAL WEIGHTED FACIAL PARALYSIS CLASSIFICATION...... 81 TABLE XI: YANAGIHARA – A REGIONAL WEIGHTED FACIAL PARALYSIS CLASSIFICATION...... 82 TABLE XII: STENNERT – A REGIONAL FACIAL NERVE PARALYSIS CLASSIFICATION...... 83 TABLE XIII: HOUSE-BRACKMANN – A GROSS FACIAL NERVE PARALYSIS CLASSIFICATION...... 84 FIGURE 17: ROSS – NEDZELSKI – THE LATEST DESCRIBED GROSS FACIAL NERVE PARALYSIS CLASSIFICATION...... 85 FIGURE 18: MICROANATOMY OF PERIPHERAL NERVES AND THEIR VASCULAR SUPPLY...... 87 FIGURE 19: STAIN-STRESS OF PERIPHERAL NERVES...... 90 TABLE XIV: INCIDENCE OF FACIAL PARALYSIS, BOTH TEMPORARY AND PERMANENT IN THE LITERATURE ...... 96 FIGURE 20: INCIDENCE OF TEMPORARY FACIAL PARALYSIS IN THE LITERATURE ...... 97 FIGURE 21: INCIDENCE OF PERMANENT FACIAL PARALYSIS IN THE LITERATURE ...... 98 FIGURE 22: FIRST PAGE OF DUPHENIX'S 1757 MANUSCRIPT ...... 102 FIGURE 23: AUTONOMIC INNERVATION OF THE PAROTID GLAND AND FACIAL SKIN – OVERALL VIEW ...... 107 FIGURE 24: AUTONOMIC INNERVATION OF THE PAROTID GLAND AND FACIAL SKIN - DETAIL ...... 109 FIGURE 25: SCHEMATIC REPRESENTATION OF THE ABERRANT REGENERATION THEORY ...... 112 FIGURE 26: FREQUENCY OF FREY SYNDROME FOLLOWING PAROTIDECTOMY ...... 116 FIGURE 27: SEVERITY OF GUSTATORY SWEATING SYMPTOMS ...... 117 FIGURE 28: MINOR-TEST SCORES IN DIFFERENT PAROTID SURGERY GROUPS ...... 118 FIGURE 29: INCIDENCE OF FREY SYNDROME IN THE LITERATURE ...... 119

4 TABLE XV: INCIDENCE OF FREY SYNDROME IN THE LITERATURE USING OBJECTIVE TESTING ...... 120 TABLE XVI: GUIDELINES FOR TOPICAL TREATMENT OF FREY SYNDROME WITH TOPICAL ANTICHOLINERGICS ...... 123 TABLE XVII: EFFECT OF VARIOUS SURGICAL TECHNIQUES FOR PREVENTING FREY SYNDROME ...... 126 FIGURE 30: PAROTIDECTOMY INCISIONS ...... 129 TABLE XVIII: EFFECT OF FIBRIN GLUE ON POSTPAROTIDECTOMY WOUND COMPLICATION ...... 132 FIGURE 31: DISTRIBUTION OF THE FACIAL INNERVATION BY THE TRIGEMINAL NERVE ...... 133 FIGURE 32: AVERAGE AND STANDARD DEVIATION OF POSTPAROTIDECTOMY RECURRENCE ...... 137 FIGURE 33: VIDEOMIMICOGRAPHY SET UP...... 145 FIGURE 34: VIDEOMIMICOGRAPHY FACIAL LANDMARKS...... 147 FIGURE 35: ADJUSTMENT OF LANDMARKS LS AND LI FOR SMILING MOVEMENT ...... 148 FIGURE 36: VIDEOMIMICOGRAPHY DISTANCE MEASUREMENTS...... 150 FIGURE 37: VIDEOMIMICOGRAPHY SURFACE MEASUREMENTS...... 151 FIGURE 38: PREFORMED STENCIL FOR FREY SYNDROME EVALUATION...... 152 FIGURE 39: STANDARDIZED FOLDING OF THE BLOTTING PAPER STENCIL...... 153 FIGURE 40: IODINE-SUBLIMATED PAPER STENCIL WETTED WITH WATER...... 154 FIGURE 41: ISPH DATA SHEET OF A PATIENT WITH FREY SYNDROME ...... 156 TABLE XIX: PERCENT CHANGE OF ALL MEASURES FOR EACH MOVEMENT IN NORMAL SUBJECTS...... 157 FIGURE 42: PERCENT CHANGE FOR EYE CLOSURE IN NORMAL SUBJECTS ...... 158 FIGURE 43: PERCENT CHANGE FOR FOREHEAD LIFTING IN NORMAL SUBJECTS ...... 158 FIGURE 44: PERCENT CHANGE FOR NOSE WRINKLING IN NORMAL SUBJECTS ...... 159 FIGURE 45 PERCENT CHANGE FOR LIP PUCKERING IN NORMAL SUBJECTS ...... 159 FIGURE 46: PERCENT CHANGE FOR SMILING IN NORMAL SUBJECTS ...... 160 TABLE XX: ANOVA OF SIDE, DAY, SUBJECT AND REPEAT VARIABLE ...... 161 TABLE XXI: PERCENT CHANGE OF THE BEST MEASURES FOR EACH FACIAL MOVEMENT WITHIN HB GRADE ...... 162 TABLE XXII: ANOVA ANALYSIS OF THE TOTAL VARIABILITY OF THE BEST MEASURE FOR EACH FACIAL MOVEMENT AND THE GLOBAL FACIAL VALUES...... 162 FIGURE 47: BOX-PLOT OF THE BEST MEASURES FOR EACH MOVEMENT AGAINST THE HB GRADE...... 162 FIGURE 48: BOX-PLOT OF THE HB GRADE AGAINST VMGS (TOP) AND VMGI (BOTTOM)...... 164 TABLE XXIII: NORMATIVE DATA OF GUSTATORY SWEATING ...... 165 FIGURE 49: NORMATIVE DATA FOR THE BLOTTING PAPER AND THE ISPH METHODS ...... 166 FIGURE 50: HISTOGRAM OF AGE DISTRIBUTION ...... 167 FIGURE 51: HISTOGRAM OF THE SIZE OF PAROTID LESIONS ...... 168 TABLE XXIV: HISTOLOGY OF PAROTIDECTOMY SPECIMEN ...... 169 TABLE XXV: FACIAL FUNCTION SCORES OF THE ENTIRE POPULATION ...... 170 FIGURE 52: POSTOPERATIVE HB SCORES ACCORDING TO WHETHER FACIAL NERVE BRANCHES WERE SECTIONED ...... 171 FIGURE 53: X-Y PLOT OF PATIENT'S AGE VS. POSTOPERATIVE HB SCORE ...... 172 FIGURE 54: BAR CHART OF THE POSTOPERATIVE HB SCORES ACCORDING TO THE TYPE OF PAROTIDECTOMY ...... 173 FIGURE 55: POSTOPERATIVE HB SCORES ACCORDING TO THE HISTOPATHOLOGY ...... 174 FIGURE 56: POSTOPERATIVE HB SCORES ACCORDING TO THREE HISTOPATHOLOGIC GROUPS ...... 175 FIGURE 57: POSTOPERATIVE HB SCORES ACCORDING TO THE SIZE OF THE LESION ...... 176 FIGURE 58: X-Y PLOT OF THE RELATION BETWEEN DURATION OF SURGERY AND POSTOPERATIVE HB SCORES ...... 177 FIGURE 59: POSTOPERATIVE HB SCORES ACCORDING TO THE TYPE OF MONITOR...... 178 TABLE XXVI: MAIN CHARACTERISTICS IN PATIENTS WITH POSTOPERATIVE HB SCORE > 2 ...... 178 TABLE XXVII: STATISTICAL RELATIONS BETWEEN PAROTIDECTOMY VARIABLES AND VMG INDEX ...... 179 FIGURE 60: DISTRIBUTION OF THE GRADES OF THE CLINICAL FREY SYNDROME EVALUATION ...... 180 FIGURE 61: OBJECTIVE FREY SYNDROME EVALUATION ...... 181 FIGURE 62: SURFACE OF FREY SYNDROME USING THE ISPH METHOD ...... 182 TABLE XXVIII: FREQUENCY OF THE DIFFERENT WOUND COLLECTION COMPLICATIONS AND TYPE OF IMPLANT ...... 183 FIGURE 63: INCIDENCE OF POSTPAROTIDECTOMY FISTULA AND TYPE OF IMPLANT ...... 184 FIGURE 64: INCIDENCE OF PAROTID WOUND COMPLICATIONS (OVERALL) AND TYPE OF IMPLANT ...... 184

5 FIGURE 65: FREY SYNDROME QUANTITY BEFORE AND AFTER TREATMENT WITH BOTULINUM TOXIN ...... 186 FIGURE 66: FREY SYNDROME SURFACE BEFORE AND AFTER TREATMENT WITH BOTULINUM TOXIN ...... 186 TABLE XXIX: CHARACTERISTICS OF THE BOTULINUM TOXIN TYPE A DILUTION AND DOSES IN PUBLICATIONS ON FREY SYNDROME TREATMENT WITH BOTULINUM TOXIN ...... 201

6 ACKNOWLEDGEMENTS

I would like to thank:

The patients, subjects of this thesis, for their cooperation and patience during the various tests for which they did not always see the importance. This study has probably improved the care they received, but future patients will probably benefit the most.

Dr D. Wang, who was a visiting Research Fellow from Fuyong University, China. He worked very hard, and with remarkable patience, to perform the majority of the computer analysis of the data collected for the videomimicography test. He also suggested the use of the global indexes for this test and essentially made a old dream of mine a realty.

Mr. G. Rizzo for his help with the video recordings during the videomimicography. These recordings scheduled at various unpredictable times often disrupted his schedule. I would also like to thank him for his help and cooperation in the setup of the computer analysis of facial landmarks, and for scanning the iodinated paper stencils used in the Frey syndrome evaluation. Without his cooperation, this work would have taken much more time to complete.

Mrs. L. Johnson, Head Nurse of the ENT Operating Room for her encouragement and interest in this study. Without her cooperation and efficacy, the purchase of the EMG intraoperative monitoring system would probably have been delayed for months in the administrative bureaucracy of our institution.

Professor W. Lehmann, Chief of the Head and Neck Surgery Division, for his enthusiasm and encouragement during this study. In large part, the ideas presented in this thesis were born and perfected during our discussions. In addition, a large number of the patients were operated on by him, and his surgical skill is reflected in the excellent results obtained.

Dr. D. Quinodoz, who generously participated in the elaboration of the functional sweat tests and who collected the patient data for the treatment of Frey’s syndrome patients with botulinum toxin.

Mr. G. Cosenday, engineer with the Geneva Cochlear Implant Center, for developing the histogram algorithms necessary for the iodine-sublimated paper method. His rapidity in providing last minute results was especially appreciated.

7 Dr. A. Vaezi who was able to propose within a very short timeframe, solutions to the reproducibility problems, which interfered with the blotting paper method. He actively participated in standardizing the perspiration tests.

Dr. P. Piletta and Dr. A. Arechalde from the Dermatology Clinic for preparing the iodinated paper stencils, as well as the temperature and skin coloration tests which were used for the objective evaluation of Frey’s syndrome.

Mr. A. Rohr who replaced Mr. Rizzo, sometimes at the last minute, in various activities with his typical eagerness. He was also responsible for the modification of the VMG chair.

Dr. D. Salomon, “Medicin-Adjoint” of the Dermatology Clinic and Dr. M. Pelizzone, “Maître d’Enseignement and Recherche” at the Geneva Cochlear Implant Center, whose ideas contributed to the initial steps which led to the new objective tests of Frey’s syndrome.

Professor P. Montandon, Dr. J. P. Guyot and Dr. S. Auderson who kindly allowed their patients to be included in this study.

Mrs. J. Newsom, my mother-in-law, whose editorial experience and professional expertise clarified several passages and greatly improved the English text

8 SUMMARY

OBJECTIVE: To review the literature on postparotidectomy complications (facial nerve paralysis, Frey syndrome, "wound complications", and recurrence) and to obtain meaningful data on their incidence, physiopathology, and favoring factors. To analyze the incidence of postparotidectomy complications in a prospective trial. To develop new objective evaluation methods for facial motor function and Frey syndrome.

METHODS: Seventy patients underwent parotidectomy between April 1994 and 1998. Several improvements of standard parotidectomy were introduced in order to decrease postparotidectomy complications.

1) Videomimicography. A new objective facial neuromuscular evaluation method, called videomimicography (VMG) is proposed. VMG uses landmarks placed on the face, and digital video recording of subjects during five standard facial movements (forehead wrinkling, eye closure, nose wrinkling, lip puckering, and smiling) performed with maximal strength. Digital video frames are directly fed in a computer and a custom-modified public domain software used for analysis. In five normal subjects, the "best measures" for each movement were assessed. The reproducibility of VMG was studied with ANOVA comparing intersubject, side, and same or different day test-retest variability. Two global values of facial paralysis were derived from these measures: the VMG score and the VMG index. The "best measures" and the VMG score and VMG index were then correlated with the facial paralysis House-Brackmann grade in 29 patients with facial paralysis.

2) Facial nerve. An intraoperative facial nerve monitoring was used in all patients. Two devices were used: a custom mechanical transducer (35 patients - 1994-6) and a commercial EMG- based apparatus (35 patients - 1996-8). All patients were analyzed, including those with cancer and those with deliberate or accidental sectioning of facial nerve branches. The outcome variables studied were the motor facial function at 1 week postoperatively (temporary paralysis) and at 6-12 months (definitive paralysis). The facial nerve was evaluated according to the House-Brackmann grading scale (HB) and according to VMG.

3) Frey syndrome. Implants were placed under the skin at the end of parotidectomy. The choice of the implant was left to the individual surgeon: 24 patients had no implant and 46 patients had the following implants – 7 lyophilized dura, 7 Ethisorb®, 32 e-PTFE sheets. The incidence of

9 Frey syndrome was evaluated 12 months after the operation. A clinical Frey syndrome was present if the patients complained of gustatory sweating or flushing. An objective Frey syndrome was present if patients tested positive with the two newly developed tests for the evaluation of Frey syndrome. The blotting-paper technique (BP) used the difference in weight of a blotting stencil before and after gustatory stimulation to measure sweating quantity. The iodine-sublimated paper (ISPH) is based on the change of color of an iodine-sublimated paper stencil by the reduction of an iodine-starch mixture by water. The stencils were digitized and the darkness was used as a measure of Frey syndrome surface as well as an important topographic indicator. The gustatory stimulus for all patients was a slice of lemon sucked for 1 minute. Some patients with an important and disabling Frey syndrome were offered treatment with an intradermal injection with botulinum toxin. The results of this treatment were evaluated by the BP and ISPH techniques before and 2 weeks after treatment.

4) Wound complications. The following wound complications were evaluated after parotidectomy: hematoma, seroma, and salivary fistula. Patients were examined daily when in the hospital and at each postoperative visit.

5) Recurrence. All patients were examined clinically at 12 months for a possible recurrence. In addition, the charts were reviewed and patients called to assess a possible long-term recurrence.

The following variables were also noted: patients sex and age, the site of the lesion, the type of parotidectomy (superficial or total) performed, the possible section of a facial nerve branch, the duration of the procedure, the histopathology and size of the lesion. Their association with the outcome variables discussed above was examined with the appropriate statistical tests.

RESULTS:

1) Videomimicography. Surface areas, close to the moving portion of the face, were found to better evaluate each facial movement studied. The reproducibility of VMG was excellent, with 80- 90% of the total variability due to intersubject variability, while the facial side, and same or different day test-retest variability was small. The "best measures" and the VMG score and VMG index had a linear and highly statistically significant correlation with the House-Brackmann scale.

2) Facial nerve. The overall incidence of facial paralysis was 27% for temporary and 4% for permanent deficits. Most of the deficits were partial, and most often concerned the marginal mandibular branch. Temporary deficits with HB>2 scores were only present in patients with

10 parotid cancer or infection. Permanent deficits were present in 3 patients, all with cancer or infection. Factors significantly associated with an increased incidence of temporary facial paralysis include the extent of parotidectomy, the intraoperative sectioning of facial nerve branches, the histopathology of the lesion, and the duration of the operation.

3) Frey syndrome. Clinical Frey syndrome was present in 12 patients: 11 without implant (53%) and 1 with implant (2.6%). Objective tests were positive in 24 patients: 16 without implants (76%) and 8 with implants (20%). In the implanted patients, the objective tests were positive in 71% of lyophilized dura, 14% of Ethisorb®, and 8% of e-PTFE patients. In all 16 patients treated with botulinum toxin, an excellent result was achieved with disappearance of clinical symptoms and an important reduction on the objective tests.

4) Wound complications. Hematoma, seroma, and salivary fistula were present in respectively 7%, 6%, and 21% of patients. Salivary fistula were more frequent with Ethisorb® (57%) and e-PTFE (25%).

5) Recurrence. 4 recurrences were noted, 2 (3.6%) in patients with pleomorphic adenoma, and 2 in patients with cancer (25%).

CONCLUSIONS:

1) The videomimicography is a promising evaluation method of facial paralysis. The test is quantitative, objective, reproducible, and rather simple.

2) The routine use of a facial nerve monitoring device was found extremely helpful during parotid surgery. In this study of unselected patients, the overall incidence of facial paralysis is 27% for temporary deficits and 4% for long-term deficits. Important temporary facial nerve deficits (HB>2) were not found in patients undergoing parotidectomy for benign tumors. Permanent deficits were present only in patients who had a section of nerve branches. Factors associated with an increased incidence of temporary facial paralysis include the extent of parotidectomy, the intraoperative sectioning of facial nerve branches, the histopathology and the size of the lesion, and the duration of the operation. A review of the physiopathological factors possibly responsible for facial nerve deficit points to nerve stretching as the most probable etiology.

3) The iodine-sublimated paper histogram (ISPH) test for facial gustatory sweating is an accurate, reliable, easy to perform, and well-tolerated objective test. The topographic information is essential for Frey syndrome treatment involving local application of a medication. In addition, the

11 quantitative data provided are indispensable to evaluate the results of a given treatment. The incidence of clinical Frey syndrome after parotidectomy is 40-50%. When objective tests are used (ISPH), the incidence is around 80%. The use of an implant placed in the wound as a prevention barrier reduces the incidence of clinical Frey syndrome to 2-3%. When objective tests (ISPH) are used, the incidence with e-PTFE is reduced to 10%. The best Frey syndrome prevention barrier appears to be a non-resorbable implant.

4) Some of the implants used (mainly Ethisorb®, but also e-PTFE) result in a high incidence of parotid fistula. Therefore, the search for the best implant should continue.

5) The treatment of Frey syndrome by intradermal injection of botulinum toxin type A appears simple, effective, reliable, fast, and devoid of major side effects.

12 1. INTRODUCTION

Parotide est une tumeur contre nature, occupant les glandules et parties d'autour, qui sont sous les oreilles dites Emonctoires du cerveau : lesquelles, parce qu'elles sont laxes et rares, facilement reçoivent les excremens d'iceluy. Ambroise Paré, 1572

The first reference to parotid pathology is attributed to Hippocrates (460-370 BC) who described purulent and non-purulent parotitis [212; 390]. Any further understanding of parotid pathology is hampered until the anatomical description of the gland and its role in secretion of saliva. In fact, as pointed out by Hyrtl in the early 19th century, even the term parotid designated a disease and not an anatomical structure [212].

The recognition of the existence and role of the major salivary glands dates back to the second half of the 17th century. Thomas Wharton (1614-1673) in a monograph named "Adenographia. Siva glandularum totius corporis descriptio" gives the first description of the submandibular gland and its drainage duct that still bears his name [189; 212; 390; 546]. At the same period, Niels Steenson also known as Nicolaus Stenonius (1638-1686) discovers the parotid duct and describes the anatomy of parotid gland [189; 212; 390; 496]. It is interesting to note, that Stenonius, after numerous contributions to medicine, embraced Catholicism and was made apostolic vicar. In 1988, he was beatified by Pope Jean Paul II [127]. During the 17th century, Malpighi studies the histology of salivary glands and coins the term "acini" [189].

Definitive demonstration of the control of salivary secretion by nerves is provided in 1850 by Carl Ludwig by stimulating the chorda tympani nerve [189]. In the second half of the 19th century, the role of the sympathetic and parasympathetic systems in salivary secretion was studied by Claude Bernard and John Langley [189; 358].

13 1.1. History of parotid surgery

L'ablation de la parotide est une opération grave, d'exécution difficile, dangereuse à cause de l'hémorragie qui l'accompagne lorsqu'elle n'est pas faite méthodiquement, entraînant fatalement une paralysie faciale définitive Charles Lenormant, 1919 [312].

The first parotidectomy was apparently carried out by the British surgeon John Hunter in 1785 for a large tumor weighting 4 kilograms [358; 390]. Despite the heroic conditions in which these operations were carried out, several surgeons have undertaken parotidectomies in the first half of the 19th century: Carmichael [82] in Ireland, Bérard [43] in France, McClellan [341], Mott [366], and Sweat [506] in the United States. In 1860 already, Brainard [58] reviewed 91 parotidectomies from 64 surgeons. Still numerous physicians doubted the feasibility of parotidectomy and McCoy wrote in 1844: “ I need to say nothing about the operation of the parotid gland, for if you consider its anatomical relation you will be convinced that its removal is quite beyond the power of surgery” [390].

In most operations on the parotid gland in the 19th century the principal problem was bleeding and the main aim, the survival of the patient. The surgical technique became established and J.-L. Faure [165] described 8 pedicles that have to be ligated and sectioned to free the gland and control bleeding. Pedicle number 5 is called the stylomastoid pedicle and is composed mainly of the facial nerve! These parotidectomies were often radical operations with removal of a various portions of the mandibular ramus, in order to completely resect the deep portion of the gland [165].

Surgical treatment of parotid tumors would be straightforward, by removing the entire gland, if not for the presence of the facial nerve in the middle of the glandular parenchyma. The possibility of accomplishing a parotidectomy without sacrificing the facial nerve was first suggested by Thomas Carwardine in 1907 [84], followed by Duval [149], Barbat [24], Sistrunk [481], and Adson and Ott [3]. The French surgical school with Duval and Redon contributed greatly to the techniques of facial nerve identification and preservation. As early as 1914, Duval [149] performs the identification and preservation of the upper facial nerve branches in benign parotid tumors by removing the inferior aspect of the mastoid process and identifying the nerve before its division. In 1916, Barbat [24] described one case in which the marginal mandibular branch is identified and followed to identify the facial nerve trunk, before removing the tumor. This report is rarely cited

14 [115], but an identical technique was popularized by Sistrunk in 1921 [481] and Adson and Ott in 1923 [3], to who credit is sometimes given [16; 79; 252; 276; 335; 527]. Sistrunk used enucleation for the majority of his cases, but stated that "In operating on some of the recurring growths and the larger growths that are deeply placed in the substance of the gland I have, in some instances, been able to save the nerve by the following procedure: I have exposed the facial nerve by first isolating the inframandibular branch of the nerve which runs along the angle of the jaw and dissecting this upward through the substance of the parotid gland to the point where the nerve divides…" [481]. The article of Adson and Ott, although published later, recommends the identification of the mandibular branch and dissection of the facial nerve in all cases [3]. This paper is a remarkable description of the surgical technique of total parotidectomy with facial nerve preservation [3]. A case report using this technique was published by Saltstein in 1936 [455].

The debate for the following 40 years will center on mixed tumors, probably because they represent more than 60% of parotid masses (see §1.3.). Because of the high number of recurrences in early publications [5; 40; 74; 281; 335; 344; 429; 494], the controversy surrounds their exact histologic nature (benign vs. malignant, uni- vs. multicentric) and, therefore, their best treatment [5; 16; 17; 35; 40; 74; 213; 241; 247; 252; 253; 276; 334; 340; 344; 407; 410; 414; 436; 493; 494; 527; 528; 545]. Benedict and Meigs in 1930 [40] reviewed 225 parotid tumors and concluded that: 1) mixed tumors are essentially benign, but recur locally with great frequency; 2) mixed tumors rarely become malignant; 3) radiation is of benefit in some benign parotid tumors, but excision is the treatment of choice. Nevertheless, the debate on each the above points carried on until the 1980's.

While the benign histology of mixed tumors became progressively established [35; 63; 78; 156; 157; 178; 274; 334; 414; 429], recurrences were attributed to inadequate surgery [34; 35; 96; 152; 252; 253; 281; 335; 492]. At that time, the majority of parotidectomies performed for mixed tumors were enucleations as described in two large series from the late 1930's [247; 407]. As late as 1940 Patey, to whom parotid surgery owes much, wrote: "There are three possible planes in which a parotid tumor can be removed. In the first place, the rather delicate capsule may be opened and the contained tumour tissue expressed… Secondly, the tumour may be enucleated in the layer of loose alveolar tissue which lies between it and the surrounding normal salivary glandular tissue… Finally, the tumour may be removed with a margin of surrounding salivary glandular tissue by cutting through or outside the gland. This procedure involves almost of necessity the cutting of some or all of the branches of the facial nerve, depending on the amount of gland removed." [407]. Papers supporting enucleation for parotid tumors have continued until the 1980's [13; 224; 232; 343; 493].

15 Bailey [16] is often credited as having described modern parotidectomy [273; 410; 527], but his technique shows minor improvements compared to Adson and Ott's 1923 description [3]. Bailey used what he called a "modified Blair" incision (see § 2.3.1.), the anterior skin flap is elevated of the gland at the beginning of the procedure, and the external carotid artery is ligated [16]. The superficial lobe is mobilized progressively starting from the inferior and anterior poles of the gland and the dissection proceeds in posterior direction "between the two lobes.” Although the facial nerve is supposed to be preserved, in Bailey's papers it is unclear, where and when it is identified. The major progress in Bailey's writing is the emphasis on complete superficial or total parotidectomy, instead of enucleation, for the surgical treatment of parotid tumors [16; 17; 248].

The first descriptions of an early identification of the facial nerve trunk followed by anterograde dissection along facial nerve branches is due to Janes, from the University of Toronto in 1940 [252]. He often removed "the tip of the mastoid process with an osteotome for easier access. The main trunk of the facial nerve is then exposed by blunt dissection as it emerges from the stylomastoid foramen… Once the nerve has been exposed it is usually surprisingly easy, except in inflammatory or malignant lesions, to dissect it free from the surrounding tissue, identifying the branches as they arise from the main trunk." [252]. Although the dates of the publications show beyond any doubt that Janes was first to describe the technique, Janes in a later article [253], stated, with great modesty, that the first parotidectomies with identification of facial nerve trunk and anterograde dissection of facial nerve branches was done for the first time around 1935 by himself in Canada, Redon in Paris, and Bailey in England!

Redon in 1945 [435], Marshall and Miles in 1947 [334], Clausen and Henley [96] in 1948, Klopp and Winship [276], and Brown et al. [63] in 1950, Hayes Martin [335] and Louis Byars [79] in 1952, described a similar technique, except for the removal of the mastoid tip. After separating the parotid gland from the sternocleidomastoid muscle [276; 435] and then from the digastric muscle [63; 335; 435], the facial nerve trunk is identified as it exits the stylomastoid foramen and dissected forward from the parotid gland. A superficial conservative (as far as the facial nerve is concerned) parotidectomy is then performed. If necessary, the deep lobe can also be removed, accomplishing what is variably called a total parotidectomy or a subtotal parotidectomy [276]. The technique was rapidly accepted by others such as Finochietto [173], Fitzgibbon [175], Brintall et al. [59], Kidd [273], Patey [410], Utendorfer [527], Beahrs [33; 34]. Although some [63; 272; 273; 276; 315; 335; 527] refer to the 1941 publication of Bailey [16], and sometimes to Janes' article [34; 59; 276; 315; 334; 492; 527], rarely is Janes credited as a pioneer. For example, Patey in his 1968 publication

16 [410] giving credit to the "three big man of parotid surgery,” cites McFarland, Bailey, and Redon, without any mention of Janes.

The anterograde facial nerve dissection technique has become the standard approach in parotid surgery [437; 520], although some surgeons still perform retrograde facial nerve dissections [93; 232; 535], following the description of this procedure by State in 1949 [492].

Janes [252] and, later, State [492] and Redon [436] condemned parotid biopsy and enucleation. Basing his observation on the work of Delarue [120], Redon [435; 436] strongly believed in the existence of multicentric mixed tumors, an idea advanced by McFarland [347; 348] and Patey [407]. This theory made him favor total parotidectomy as the routine surgical treatment of mixed tumors, an attitude still followed by French surgeons [6; 94; 211; 270; 288; 291; 520].

Great progress in the comprehension of mixed tumors was made by the serial sectioning of parotidectomy material by Patey and Thackray [414]: 1) the capsule surrounding the tumors was found to vary greatly in thickness, not only from case to case, but also in different areas of the same tumor; 2) the capsule was found sometimes to be incomplete with tumor adjacent to normal gland tissue; 3) mixed tumors were found to be nodular and to grow by polypoid extensions through the capsule; 4) these polypoid extensions were more common in small tumors; 5) no multifocal mixed tumors were found in 37 specimens [414]. After recurrence, mixed tumors were often multicentric in the entire operated field and in particular along the previous scar [414]. These conclusions have been confirmed by Gunnel [214], Kleinsasser [275], Naeim et al. [372], Danovan and Conley [113], Lawson [305], and Lam et al. [299]. Also, Patey and Thackray's finding explained McFarland's conclusion from the 1930's, that small mixed tumors recur more often than large ones, and therefore that small tumors should not be operated on [347]. Because of McFarland's position as a prominent pathologist with special expertise in parotid tumors [344; 345; 346; 347; 348], the above statement haunted parotid surgeons [16; 17; 63; 253; 334; 407] until Patey and Thackray's study published in the late 1950's.

Since mixed tumors are benign, recurrences are often delayed, up to 10-20 years after the initial surgery. Therefore, after the progressive generalization of superficial parotidectomy as minimal procedure in the 1950's, series with low recurrence rates for mixed tumors appeared [35; 63; 205; 253; 273; 527]. Also, comparison between enucleation with superficial or total parotidectomy showed lower rates than enucleation [35; 205; 253; 530] and have continued to the 1980's [104; 493; 498].

17 In view of the high recurrence rates of mixed tumors in earlier series, radiation therapy has been proposed as adjuvant therapy since the 1920's [74], either as external beam radiation [5; 13; 224; 247; 343; 407; 456; 528], curietherapy with radium implants [74; 343; 528], or an association of both [5; 247]. Although the available results do not show a reduction of recurrence rates by radiation [528] and several authors have recommended against its use [16; 17; 152; 253; 273; 347; 407], proponents of radiation have continued until the 1990's [13; 25; 194; 224; 315; 343; 456; 528].

On a somewhat different front, the progress in the treatment of malignant parotid tumors has been made by a more precise and adequate histopathological diagnosis [29; 30; 31; 32; 178; 478; 515], by the recognition that recurrence and survival have to be measured over a time period of 10- 20 years [47; 159], and by the elaboration of groups of malignancies [47; 261] requiring treatment approaches of increased aggressiveness. However, as already noted by Brown et al. [63] in 1950, in view of the difficulty to establish an exact pre- or intraoperative pathological diagnosis in parotid malignancies, and because of the lack of an adequate rehabilitation techniques in facial nerve paralysis [199; 451], the often used approach is to make every effort to conserve a nerve not paralyzed prior to surgery [409].

In summary, the progress in parotid surgery has been a better understanding of mixed tumors and a standardization of surgical procedures towards early identification and dissection of the facial nerve. However, some of the problems discussed in the 1930's are still not completely resolved today. While it is now generally accepted that pleomorphic adenomas are benign, the extent of primary surgery for pleomorphic adenoma is still debated [132]. Similarly, radiation for pleomorphic adenoma, after a long debate, is less often used but not abandoned. In addition, the frequency and importance of various surgical complications of parotidectomy are reported over a wide range, and few techniques have evolved in order to reduce them [135; 137]. Finally, if the preservation of the facial nerve in benign parotid tumors has become the standard of care, its handling in the various parotid malignancies is still debatable.

18 1.2. Anatomy It follows from its anatomy and complex relations of the parotid, that its entire removal as a surgical procedure is an anatomical impossibility Treves, 1907[519]

The parotid gland fills the parotid space. The exact three-dimensional anatomy of the parotid space is extremely complicated and variable, but, with simplification, has been schematized as a quadrangular pyramid which is upside-down and the summit of which is somewhat truncated [56] (Figure 1). Therefore, six different walls can be recognized to the parotid space [33].

Figure 1: Schematic representation of the parotid gland as a truncated quadrangular pyramid. Four of the six walls are represented: Ant: anterior wall, Sup: superior wall, Post: posterior wall, Inf: inferior wall. The front of the figure represents the lateral wall and the back the medial wall.

19 The anterior wall is made of the mandible and the adjoining masseter and internal pterygoid muscles. The parotid gland actually extends more superficially than the masseter in the direction of Stenson's duct. The proximity of these muscles to the parotid gland can explain the frequently encountered post-operative complaint of painful chewing. The limit between the anterior and medial wall is the sphenomandibular ligament, which is actually a thickening of the interpterygoid aponeurosis [424].

The medial wall is quite complex: here the parotid is bound by the deep parotid aponeurosis, which is attached to the sphenomandibular ligament and the stylomandibular ligament. Between the mandible and the spheno-mandibular ligament is a potential space called the retro-condylar space of Juvara [266] (also called the stylomandibular tunnel by Patey [413]), through which the auriculotemporal nerve and the internal maxillary artery enter/leave the parotid space [424]. Deep- lobe parotid tumors egress out of the parotid space and extend into the parapharyngeal space, through this retro-condylar space. As these tumors grow, they are blocked and assume a dumb-bell shape, a term coined by Patey [413]. Another weak spot, allowing for communication between the parotid and parapharyngeal spaces, is present between these ligaments.

The posterior wall is made of: 1) the styloid process and its muscles and ligaments, namely from medial to lateral: the stylopharyngeus muscle, the styloglossus muscle, the stylohyoid ligament and the stylohyoid muscle; 2) the posterior belly of the digastric muscle and; 3) the sternocleidomastoid muscle. The styloid muscles and ligaments form the so-called styloid "diaphragm" [83; 424; 514], which separates the parotid space from the carotid sheath. The facial nerve enters the parotid space between the digastric and stylohyoid muscle [424].

The extreme irregularity of the posterior and deep surfaces of the parotid space, and the absence of an obvious angle between them, has led to alternative descriptions of the shape of the parotid space [260]. A single posterior (in fact posteromedial) surface is described, along with the usual anterior and superior surfaces. The inferior surface is seen as an edge, and as a result, the gland is described as a triangular pyramid [33].

The superior wall has, actually, more of a triangular shape [256]. It is formed laterally by the zygomatic arch and by the cartilage of the external auditory canal. More medially, the upper limit of the parotid space is formed by the skull base, which is made, at this level by the tympanic portion of the temporal bone. The glenoid fossa of the temporal is just in front of this extension, so it is not surprising that it could be invaded in certain cases of parotid cancer [442].

20 The inferior wall is formed, posteriorly, by the convergence of the stylohyoid ligament and muscle with the digastric tendon at the superior edge of the hyoid bone. The anterior edge is the mandibular angle with the attachment of the stylomandibular ligament. Between these two bony landmarks is a fascial thickening, named the angular tract of the cervical fascia [192; 565], separating the parotid space from the submandibular space. This fascial thickening extends sometimes to the sternocleidomastoid muscle forming the so-called sternomandibular ligament [424; 514]. The lowest point of the parotid gland is 1.18 ± 0.51 cm below and 1.38 ± 0.32 cm behind the angle of the mandible [566].

The lateral wall is apparently simpler, since it is composed only of aponeurosis and skin. Nevertheless, considerable debate exists on the exact nature of the aponeurotic coverage of the parotid and the arrangement of fascia in the preauricular region (see §1.2.1.).

The height of the parotid space has been measured to average 5.8 cm by Davis et al. [115] and 4.2 ± 0.24 by Rudolph [452], while the width was estimated at 3.4 cm by Davis et al. [115] and 2.43 ± 0.31 cm by Rudolph [452].

21

Figure 2: Schematic representation of the parotid gland. The drawing represents a section through the parotid gland and surrounding structures. The buccopharyngeal fascia (1) covers the lateral aspect of the pharynx represented by the superior constrictor muscle (17) and tonsil (15). The masseteric fascia (10) covers the internal pterygoid muscle (14), the masseter and encircles the buccal fat pad of Bichat (12). The deep parotid fascia (7) covers the medial and deep aspect of the gland. On the medial aspect the sphenomandibular (17) and stylomandibular ligament (2) are shown. The posterior wall is made of the styloid muscles and ligaments: stylohyoid ligament (3), stylohyoid muscle (4), as well as the digastric muscle (5). It continues as the superficial layer of the deep cervical fascia, covering the sternocleidomastoid muscle (6). The external carotid artery (8) and posterior facial vein (9) are seen in the deep aspect of the gland. The Stenson's duct (11) is seen passing forward from the gland to terminate by piercing the buccinator muscle (13). Reprinted from Proctor [424], without permission.

22

Figure 3: Horizontal section through the neck at the level of the parotid gland. 1) Maxilla – symphysis; 2) Orbicularis oris muscle; 3) + 15) Stenson's duct; 4) Platysma muscle; 5) Bichat's fat pad; 6) Masseter muscle; 7)+ 19) Mandible – ascending ramus; 8) Internal pterygoid muscle; 9) Cartilage of the external ear; 10) Styloid process; 11) + 22) Internal jugular vein; 12) + 24) Mastoid process; 13) Cerebellum; 14) Tongue; 16) Mandible – body; 17) Oropharynx; 18) Internal maxillary artery; 20) Internal carotid artery; 21) Retromandibular vein; 22) Vagus, glossopharyngeus, spinal, and hypoglossus nerves. P = parotid gland.

23 1.2.1. Parotid fascia

The fascias in the human trunk are divided into a superficial and a deep fascial layer [209; 514]. The superficial fascia is under the dermis and is usually a thin connective tissue layer. The superficial fascia extends from the vertex where it covers the epicranius muscle, continuing on to the face, neck, chest, and abdomen [209]. As a rule, vessels and nerves are located deep to the superficial fascia, except fine elements destined to the overlaying skin. The deep fascia also extends from the abdomen to the head.

In the neck, the superficial cervical fascia covers the superficial aspect of the platysma muscle and, according to Grodinsky and Holyoke, splits to also cover its deep surface [209]. Therefore, the superficial fascia constitutes a complete subcutaneous layer surrounding the entire circumference of the neck. It is attached to overlaying skin within thin fibrous septa and continues on the face and scalp.

In the neck, the deep cervical fascia has a complicated structure and, unfortunately, the names chosen render the subject confusing and misleading. Grodinsky and Holyoke divide the deep cervical fascia in 3 layers: a superficial layer of the deep cervical fascia (SL-DCF), a middle layer, and a deep layer [209]. Obviously, confusion exists when speaking about a "superficial fascia" in the neck: is it the superficial cervical fascia or the superficial layer of the deep cervical fascia. Furthermore, while the Grodinsky's terminology has become standard nomenclature in Otolaryngology – Head and Neck Surgery textbooks [206; 259; 471], it is not exactly followed in anatomical textbooks. Also, none of the anatomical descriptions done recently refer to Grodinsky and Holyoke's work [204; 265; 361; 502; 516; 567].

The deep layer of the deep cervical fascia surrounds the vertebra and the prevertebral and paravertebral musculature (Figure 4). The middle layer of the deep cervical fascia encloses the pre- laryngeal musculature, the thyroid gland, trachea, and esophagus-pharynx. This layer is most developed in the infra-hyoid portion of the neck and continues above that level as a fascial covering of the pharynx up to the base of skull, where it is called the buccopharyngeal fascia.

The SL-DCF forms, like the superficial cervical fascia, a complete aponeurotic layer surrounding the entire neck circumference. It is superficial to the pre-laryngeal musculature, splits to cover both sides of the sternocleidomastoid muscle, reunites to split again to ensheath the trapezius muscle, and continues on to the spines of the cervical vertebrae (Figure 4). The SL-DCF attaches to the hyoid and then splits to form the capsule of the submandibular gland. Further

24 superiorly and laterally, the SL-DCF attaches to the mandible and splits to encircle the masseter, mandible ramus and pterygoid muscles bounding the so-called masticatory space [209] (Figure 3 and 4). Similarly, the SL-DCF covers the deep aspect of the parotid space. The SL-DCF continues in a superior direction over the zygomatic bone and temporalis muscle as the deep temporal fascia [503].

Therefore, as far as the parotid region is concerned, the main layers involved are the superficial cervical fascia and the SL-DCF [209; 514]. The main subject of dispute is the relative contribution of the SL-DCF to the superficial parotid aponeurosis. Grodinsky and Holyoke [209], and Testut [514] favor the presence of a complete aponeurotic layer from the SL-DCF covering the parotid gland. On the other hand, Coller and Yglesias found a capsule only on the deep side of the parotid gland, the superficial side being covered only by the superficial layer [103].

Renewed interest to this apparently scholastic dispute was brought by the description of the so-called superficial musculo-aponeurotic system, or SMAS, by Mitz and Peyronie [361; 513]. The SMAS is a key concept in modern rhytidectomy [438]. Early facelift techniques were essentially skin mobilization through undermining and excision [23; 263], while second generation facelift techniques [402; 483] involve dissection of the SMAS down to the platysma and its cephaloposterior advancement. Because of the clinical relevance, the exact anatomical fascial layering and the relation to facial nerve branching has generated a large controversy. Recently, the SMAS technique has been claimed to provide better postparotidectomy esthetic results [7; 264; 428] and to prevent Frey syndrome [7; 51; 428; 562].

According to Mitz and Peyronie [361], the SMAS: 1) is a continuation of the superficial cervical fascia - platysma complex; 2) attaches anteriorly to the muscles of facial expression; 3) separates the subcutaneous fat in two layers; 4) is independent of the "parotid fascia" which is located deep to the SMAS. On the other hand, Jost and Levet [265] show different facial layers: 1) a superficial fascia under the skin; 2) a layer continuing the platysma and covering the parotid gland, which is called the "parotid fascia". The "deep fascia" (no precise reference to a corresponding neck structure) is located only on the deep aspect of the parotid gland (Figure 5 and 6).

A lack of generally accepted agreement persists [438; 537]. It is now accepted that the SMAS layer is a continuation of the platysma neck layer [163; 204; 438; 502; 516; 537]. What is disputed is whether there is a distinct superficial fascial layer lateral to the SMAS [265; 516], whether the parotid fascia is a separate layer deep to the SMAS [163; 361; 502], or whether a single subcutaneous layer is present [537]. The more recent description by Gosain et al. [204] might bring

25 a consensus: 1) laterally to the SMAS the fibrous septa joining the superficial fascia to the skin are well developed over the parotid and could have been interpreted by Jost and Levet as a separate fascia; 2) the SMAS is below the fat and represents a continuation of the platysma - superficial cervical fascia; 3) a distinct parotid fascia is present covering the gland, but it is extremely thin, explaining why it could have been missed by previous investigators. In addition, significant individual variation is present in the thickness of these different layers [204].

Figure 4: Schematic representation of the fascial layers of the neck The three layers of the superficial layer of the deep cervical fascia are drawn: in red the superficial layer, in blue the middle or visceral layer, and in green the deep layer.

26

Figure 5: Schematic representation of the fascial layers at the level of the hard palate The facial extension of the superficial layer of the deep cervical fascia is shown to surround the parotid gland and masseteric space on both superficial and deep face. Reprinted without permission from Grodinsky and Holyoke [209]. Abbreviations: MAX.S. = Maxilla sinus; MAND. = Mandible; MASS.M. = Masseter muscle; PT.INT.M. = Pterygoid internal muscle; PT.EXT.M. = Pterygoid external muscle; MASTIC.SP. = Masticator space; PAROT.GL. = Parotid gland; E.C.A. = External carotid artery; P.F.V. = Posterior facial vein; MAST.PROC. = Mastoid process; DIG.POST.M. = Posterior belly of the digastric muscle; ST.CL.M. = Sternocleidomastoid muscle; LAT.PHARYNG.SP. = Lateral pharyngeal space; TEN.VEL.PAL.M. = Tensor velopalatini muscle; LEV.VEL.PAL.M. = Levator velopalatini muscle; AUDIT.TUBE = Auditory tube; PHARYNX VISC. F. = Pharyngeal visceral fascia; ALAR F. = Alar fascia; PREVERT.F. = Prevertebral fascia; SCAL.F. = Scalene fascia; I.J.V. = Internal jugular vein; C.A. = Carotid artery.

27

Figure 6: Anatomical cross-section showing the fascial layers lateral to the parotid gland Abbreviation: M – mandible and masseter muscle, P – parotid gland, S – submandibular gland, SCM – sternocleidomastoidien, SF – superficial fascia, T – temporalis muscle, Z – zygomatic bone. The discussion is centered about the fascia layers included in the fascia marked by ?. It should contain the SMAS, which is the facial extension of the platysma and its fascia, and the facial extension of the SL-DCF. Whether this fascial layer can be divided into 2 layers, with a separate parotid fascia is debatable. The superficial fascia below the epidermis is clearly seen.

28 1.2.2. Contents of the parotid space

The majority of the parotid space is occupied by the parotid gland. Other anatomical structures (Figure 7) enclosed by the parotid fascia include a) arteries: the external carotid artery and its terminal branches; b) veins: the external jugular vein and its branches; c) lymphatic vessels and ganglions; d) nerves: the facial nerve, the great auricular nerve, and the auriculotemporal nerve.

The external carotid artery ascends behind the digastric muscle [424] to enter the parotid space between the stylohyoid muscle and ligament [424]. It travels on the posterior and middle wall of the parotid space and gives off, within the parotid space, postauricular, including stylomastoid, and parotid branches. The terminal bifurcation of the external carotid artery into superficial temporal artery and internal maxillary artery usually takes place in the parotid space [424]. The superficial temporal artery ascends lateral to the zygoma, joined by the temporal branches of the facial nerve (§1.2.5.), while the internal maxillary artery pierces the deep parotid aponeurosis, in the retrocondylar space of Juvara [424], to enter the parapharyngeal space.

The retromandibular vein (retrofacial, posterior facial vein) is formed by the union of the superficial temporal vein and the internal maxillary vein(s) at the level of the maxillary condyle [260; 297; 514]. The vein is entirely within the parotid gland and is located within 5 mm of the branching of the facial nerve [277; 297]. The postauricular and occipital veins join the retromandibular vein either within the parotid space, or just inferior to it. The retromandibular vein gives of a communicating branch for the facial vein and continues in the neck as the external jugular vein [424]. Dargent and Duroux [114] attempted to schematize the venous drainage around the parotid gland and found this so-called textbook pattern in 71%, the remaining cases having large veins in front of the facial nerve and its branches.

Numerous lymphatic ganglions are located within the parotid space. They drain the scalp and the face. During development (see §1.2.4), the parotid tissue develops before the parotid and cervical fascias [196; 260], resulting in the encapsulation of lymphatic ganglions within the parotid gland [381].

The auriculotemporal nerve enters the parotid space through the retrocondylar space of Juvara [33; 424] and travels along the medial, than posterior aspect of the gland, crossing behind the superficial temporal vessels [260]. It than passes between the gland and the bony and cartilaginous external ear canal [33; 412], to provide the sensory innervation of the facial skin in front of the

29 auricle. Parasympathetic fibers from the inferior salivary nucleus and destined to the parotid gland travel within this nerve.

The saliva secreted by the parotid gland drains in Stenson's duct, which leaves the gland and the anterior wall of the parotid space. It runs in an anterior direction on the surface of the masseter muscle. Reaching the anterior border of the muscle, the duct turns sharply medianwards, to pass through Bichat's fat pad and the buccinator muscle, piercing the oral mucosa at the level of the upper second molar (Figure 2).

1.2.3. Histology

The basic secretory unit of all salivary glands is the acinus made of a single cell row of secretory cells. The acini are enveloped by the processes of myoepithelial cells containing smooth muscle-like contractile filaments. The contraction of the myoepithelial cells forces the fluid secreted by secretory cells out of the acinus and in the duct system of the gland. Several acini empty in the so-called intercalated ducts, which also contain secretory cells in their lining. The intercalated ducts join to form larger ducts called striated ducts, named so because of their striated appearance in light microscopy. These striations result from the presence of numerous infoldings of the plasmalemma on the basal side of the cell. These numerous plasmalemma infoldings contain numerous mitochondria and are characteristic of epithelia involved in the transport of water and solutes, such as the renal distal tubules and the stria vascularis of the cochlear duct. The intercalated ducts drain into lobular duct, which join to form Stenson's duct.

The salivary glands have, anatomically, both a sympathetic and parasympathetic innervation at the level of the acinar and myoepithelial cells [179]. Richins and Kuntz have demonstrated in the cat that parotid secretion is mainly under parasympathetic cholinergic control [441]. Gustatory (via the chorda tympani or glossopharyngeus nerves), visceral (via the vagus nerve), and somatic (via the chorda tympani nerve) stimuli elicit an increased salivary secretion [510]. The role of sympathetic system is modest and probably mediated through the intraglandular vessels [180; 441].

30

Figure 7: Schematic representation of the vessels and nerves within the parotid space. The drawings represent the parotid gland partially sectioned and the vessels and nerves within the parotid space. The facial nerve (1) exits from the stylomastoid foramen and divides within the parotid gland into a temporofacial division (14), which gives temporal (22), zygomatic (21,18,17), and buccal (15) branches and a cervicofacial division (13) giving marginal mandibular (12) and cervical (10) branches. The faciovenous plane of Patey is formed by the divisions of the facial nerve and the intraparotid veins. The retromandibular vein is formed by the confluence of the superficial temporal vein (23) and the internal maxillary vein (19). After sending an intraparotid communicating vein (9) to the facial vein, the retromandibular vein continues as the external jugular vein (6). The external carotid artery (7) gives off local branches such as the postauricular artery (3) and stylomastoid artery (2) before dividing into superficial temporal artery (23) and internal maxillary artery (19). The auriculotemporal nerve (20) enters through the retrocondylar space of Juvara to pass behind the gland and the superficial temporal vessels (23) to exit in front of the external auditory canal (24). The great auricular nerve gives a parotid branch (5). Also (4) digastric muscle, (11) sternomandibular ligament, (16) transverse facial artery. Reprinted from Proctor [424], without permission.

31 1.2.4. Parotid lobes

The parotid gland was viewed as a bilobated structure, following the description by Grégoire [207] and McWhorter [354] in the earlier part of the century. For Grégoire the facial nerve is the bookmark placed in the book-like parotid gland, with the bookbinding located superiorly ("le nerf facial a l'air d'être placé entre deux couches de glande se continuant par leur extrémité supérieure, comme un signet dans un livre, dont la reliure serait tournée en haut") [207]. McWhorter found the isthmus between the divisions of the facial nerve and compared the gland to an "H" [354]. Clinicians also shared this point of view [16; 492] and Bailey wrote in 1941 that "the facial nerve is like the meat of the sandwich" [16]. Apparently, the belief in this bilobated structure helped surgeons to accept the possibility of a superficial parotidectomy [18; 83; 115; 335; 412; 492].

Anatomically the mandible and its muscles i.e. the masseter on the outside and the internal pterygoid on the inside can be seen to form the anterior border of the isthmus, while the mastoid and styloid processes can be seen to bound the isthmus posteriorly. The facial nerve and its branches are located in this vertical plane and can be seen to form a division plane within the gland. The use of this plane during parotidectomy to free the gland from the nerve and the established nomenclature of parotidectomy reinforces the two-lobe concept. Patey phrased nicely the problem: "some have concluded that the splitting of the gland into two parts represents not merely a convenient surgical technique, but an actual anatomical subdivision of the gland into two lobes, i.e., that the parotid gland is not merely surgically bipartite but anatomically bilobar" [412].

The two-lobe concept has been abandoned on several anatomical and embryological grounds. First, lobes are demarcated by fissures, sulci, connective tissue and this is not routinely observed in the parotid gland [190; 412; 552]. Second, two lobes suppose that the draining ducts will progressively unite to form a deep and a superficial duct that should join to form Stenson's duct. Again, this has not been observed histologically nor during endoscopy [332; 333], and actually numerous deep lobe collecting ducts cross the facial nerve plane to terminate in the collecting ducts of the superficial portion of the gland [190; 351; 354; 412; 552]. The relationship of the ductal branching system to the facial nerve was nicely studied by McKenzie [351] and later by Winsten and Ward [552], and Patey and Ranger [412] by making plastic casts of the drainage system of the parotid gland and comparing it to the anatomy of the facial nerve and its branches. Numerous ductal crossing were seen across the facial nerve plane [115; 351; 412; 413]. McKenzie's conclusion is eloquent: "A better analogy than a parotid sandwich would be to compare the gland to a creeper weaving itself into the meshes of a trellis-work fence" [351]. After stating that "the

32 bilobar concept has been of value in helping the development of the operation of parotidectomy with conservation of the facial nerve, but now that the operation is established there can be no justification for the persistence of a false concept", Patey [412] proposed to name the divisions of the parotid gland suprafacial and subfacial parotid.

Third, the parotid gland develops embryologically as a structure on both sides of the facial nerve [190; 552]. The majority of the superior portion of the gland develops deep to the upper division of the facial nerve, while most of the parotid parenchyma is located superficial to the lower facial nerve branches [190] (Figure 8). The mastoid process develops rather late and pushes some of the parotid gland deep to the facial nerve plane and towards the parapharyngeal space [190; 422].

Figure 8: Schema of the development of the facial nerve and parotid gland Reprinted from Gasser RF [190] without permission. The parotid primordium develops as a thickening of the buccal epithelium and extends in a posterior direction, while facial nerve branches, after exiting the temporal bone, grow anteriorly towards the muscles of facial expression. The parotid gland development is shown with dark arrows, around the facial branches in white. The width of the black arrows shows the sequence of development of the parotid primordium.

Nevertheless, this plane remains important for the actual parotid surgery. It is located slightly superficial to the plane occupied by the intraparotid veins and a so-called "fasciovenous plane" is discussed by Patey [412]. It is important to realize that this plane is neither strictly vertical, nor is the same amount of parotid tissue found superficially to the facial nerve: in the superior portion of the gland the temporal and zygomatic branches of the facial nerve are more superficial than the

33 facial nerve branches in the lower portion of the parotid [190; 192]. In addition, as the facial nerve plane is followed anteriorly it becomes more superficial, with less parotid tissue covering the nerve [424].

1.2.5. Facial nerve anatomy

The facial nerve, after exiting the stylomastoid foramen, is almost immediately surrounded by parotid gland tissue [412; 424], although others pointed that a retro-glandular portion of the nerve is present [196]. Redon [437] called this segment the useful operating length ("la longueur opératoire utile"). As stated earlier the nerve enters the parotid space between the digastric and stylohyoid muscle [424]. The nerve is lateral (more superficial) to the styloid process and to the most superior extension of the posterior belly of the digastric muscle (Figure 7). Before entering the parotid gland, branches for several muscles leave the facial nerve: the occipital belly of the epicranius muscle, the auricular muscles, the stylohyoid muscle and the posterior belly of the digastric muscle [192].

The point of entry of the facial nerve into the parotid gland has been described as the middle of a line uniting the tip of the tragus to the angle of the jaw [424]. Within the parotid gland, the facial nerve takes an arciform course, which is concave upward and medianward [115; 424]. It is in close relation to the stylomastoid artery [424], which arises from the occipital or postauricular branches of the external carotid artery and is responsible for vascularization of the mastoid segment of the facial nerve [50]. The distance between the facial nerve trunk and the closest point of the digastric muscle was found to be 9 ± 2.7 mm and the distance to the closest point of the external auditory canal was found to be 11 ± 3.4 mm [239]. The depth of the facial nerve trunk from the skin surface was measured in cadavers as 20.1 ± 3.1 mm [452].

Davis et al. [115] studied in the 1950's, 350 facial halves and described the detailed anatomy of the extratemporal facial nerve – it is a landmark article that every parotid surgeon should read. The facial nerve always branches within the parotid gland into two main divisions, named the temporofacial and cervicofacial divisions [115; 192; 268; 384; 412; 424]. The branching point is known as the pes anserinus [18; 260]. The portion of the faciovenous plane of Patey included by the two primary divisions is sometimes called the triangle of Friteau [424]. The length of the extratemporal facial nerve trunk, i.e. the distance between the stylomastoid foramen and the division has been found by Dargent and Duroux as 13 mm [114]. However, more recent

34 measurements by Holt [239] show this distance to be 21 +/- 4.1 mm. The division is located on average 6 mm posterior to the posterior border of the mandible [114]. The average distance between the pes anserinus and the angle of mandible is 3.2 cm, with a minimum of 2.5 cm and a maximum of 4.5 cm [115].

The temporofacial division was found to be larger in all instances [115; 268]. The temporofacial division has an almost horizontal direction and subdivides into temporal and zygomatic branches. The cervicofacial division has a quasi-vertical direction and subdivides into cervical and marginal mandibular branches. The buccal ramus is in between, and it can arise from either division (Table 1 and Figure 9). Davis et al. classified the plexiform division patterns of facial nerve in 6 different types with very few anastomosis between branches in type I and extensive anastomosis in type VI (Table 1 and Figure 9) [115]. More recently, Katz et Catalano presented a slightly different classification, into 5 different patterns [268]. Four of these 5 patterns are somewhat similar to the original Davis et al. classification [115], while the last one (frequency of 3%) is characterized by double main facial nerve trunks. Several generalizations can be made:

1) Branches located at the extremities of the nerve distribution receive fewer anastomosis with other branches;

2) The majority of anastomosis occur between the buccal and zygomatic divisions, forming the so-called parastenon plexus [384; 518];

3) The number of anastomosis decreases in caudal branches [115; 335], with the marginal mandibular branch receiving anastomosis in only 6.3% [115];

4) There is no anastomosis between the cervical and other branches [115];

5) The anastomosis are more extensive when the buccal branch arises from the cervicofacial division [115; 566].

These classifications are more or less objective attempts to classify the highly individual and variable bifurcation patterns of facial nerve branches. In a different and more objective approach, Schwember and Rodriguez [470] analyzed the various facial nerve branches, as they excited the anterior borders of the parotid gland, and measured the distance between these exit points and fixed facial landmarks. They did not provide a classification schema because of the important variability but concluded that by the time the nerve exited the gland, there is not 5 but between 7 and 11 different facial nerve branches [470]. This article is difficult to summarize but should be consulted when a retrograde dissection of the facial nerve is planed.

35 Since the facial nerve is often identified on preoperative CT-scan or MRI by the venous plane behind it, the relationship of the retromandibular vein to the nerve has been the subject of several studies. The retromandibular vein is formed by the union of the superficial temporal vein and the internal maxillary vein at the level of the maxillary condyle [297; 514] (see §1.2.2). The vein is entirely within the parotid gland and is located within 5 mm of the branching of the facial nerve [277; 297] with variations between 2.3 to 5.1 cm [277]. All facial nerve branches were superficial to the retromandibular vein in 90% of cases for Laing and McKerrow [297] and Kopuz et al. [277], while Dingman and Grabb found this relationship in 98% of their dissections [125]. In the remaining cases, the retromandibular vein was superficial to branches of the inferior division, but always deep to facial nerve branches arising from the upper division [277; 297].

Frequency Frequency Buccal branch Type Anastomosis between FN branches (Davis, [115]) (Katz, [268]) origin

I 13% 24% No anastomosis between facial branches Temporofacial

Between zygomatic and buccal branches (parastenon Temporofacial II 20% 15% anastomosis [518], zygomatic loop [268])

Single anastomosis between zygomatic and buccal branch. Cervicofacial III 28% The anastomosis is distal anterior to the parotid tissue. 44% Anastomosis between temporal and zygomatic branches and Cervicofacial IV 24% extensive anastomosis between zygomatic and buccal branches. Also some bucco-mandibular connections [268].

Combination of type II and III. Extensive anastomosis Cervicofacial or V 9% between zygomatic and buccal branches. both [268] 14% Plexiform anastomosis between temporal, zygomatic, buccal Cervicofacial or VI 6% and marginal mandibular rami. both [268]

VII 0% 3% Double main facial trunk Temporofacial

Table I: Schematic description of the branching patterns of the facial nerve After Davis et al. [115] and Katz and Catalano [268]. See also figure 9.

Anterior to the parotid gland, the facial nerve branches are covered by the SMAS (superficial cervical fascia) and the masseteric fascia (superficial layer of the deep cervical fascia) [27; 196]. The depth, from the overlaying skin, of various facial nerve branches at their points of exit from the parotid gland was measured by Rudolph at about 10 mm [452]. Further nerve divisions are

36 observed as branches approach the muscle they innervate, always on the deep aspect of the muscle [44; 317]. The peripheral anatomy of facial nerve branches is relevant for earlier parotidectomy techniques, for parotidectomy reoperations where the facial nerve trunk might be difficult to identify, as well as for rhytidectomy [27; 203].

1.2.5.1 Marginal mandibular branches. Dingman and Grabb [125] studied the position of the marginal mandibular branch, relative to the inferior edge of the mandible, by dissecting 100 facial cadaver halves: posteriorly to the facial artery the marginal mandibular branch was located above the inferior edge of the mandible in 81% of the specimens, while anterior to the facial artery the marginal mandibular branch was above the mandible in 100% of the specimens. Several clinical dissections studies in alive patients [20; 44; 268] and cadavers [444; 566] disagree with this assessment, because the marginal mandibular branch was often found below the lower border of the mandible. This discrepancy is yet to be explained but possible reasons include neck extension during surgery and cadaveric tissue fixation [20]. Rodel and Lang found in cadavers that the marginal mandibular nerve is below the inferior edge of the mandible in 77% of cases, with an average distance of 6.5 mm (2 – 14 mm) [444]. For Ziarah and Atkinson, the nerve was below the mandible in 53% of their cadaver specimens, and in 6% the nerve was below the mandibular for more than 1 cm anterior to the facial vessels [566].

The so called marginal mandibular branch was found to be an unique nerve in 21%, two branches in 67%, three branches in 9%, and 4 branches in 3% [125]. A similar distribution was found by Rodel and Lang [444] (1 nerve in 45%, 2 nerves in 33%, 3 nerves in 20%, and 4 nerves in 2%) and Ziarah and Atkinson [566] (1 nerve in 36%, 2 nerves in 53%, 3 nerves in 11%). This nerve leaves the parotid gland approximately 1 cm below the level of the mandibular angle [203] and its branches are located under the platysma [317; 382; 566] until about 3 cm from the corner of the mouth, where they pass between the platysma and the superficial layer of mimic muscles [317]. The marginal mandibular nerve branches innervate the mentalis, the depressor labii inferioris (triangularis), and the depressor anguli oris (quadratus) muscles [382] and possibly the upper aspect of the platysma [20].

Ziarah and Atkinson [566] studied the relationship of the marginal mandibular branches and the facial vessels: in their 110 cadavers halves, the nerve was always superficial to the facial vein, while the relationship to the facial artery was variable.

The depth of the marginal mandibular branches, 5 cm in front of their exit points from the parotid gland was found to be 4.4 ± 2.3 mm [452].

37 1.2.5.2. Cervical branches. The cervical branch travels deep to the platysma and innervates only this muscle [565]. It is supposed to be identified as it crosses lateral to the retromandibular vein [33; 93; 303]. However, this relationship is inconstant in 10% of specimen, with the inferior facial nerve branches passing deep to the retromandibular vein [277; 297]. It is located 0.83 ± 0.37 cm behind the angle of the mandible [565]. The cervical branch was found to be unique in 80% of cases, while in the remaining 20% two branches were found [565]. Division in terminal plexus occurs below and anterior to the hyoid bone [565].

1.2.5.3. Buccal branches. The peripheral location of "midfacial" (mostly buccal) branches has been described as deep to the SMAS [361], deep to the masseteric fascia and superficial to Bichat's fat pad [20], and deep to Bichat's fat pad [470]. Buccal branches are said to be identifiable superficially and parallel to Stenson's duct [260]. The depth of buccal branches, 5 cm in front of their exit points from the parotid gland was found to be 6.7 ± 2.5 mm [452].

Buccal branches innervate the buccinator and the orbicularis oris muscles. The procerus, the zygomaticus major and minor, the levator labii superioris, the levator anguli oris, and the buccinator muscles can be innervated by buccal and/or zygomatic branches.

1.2.5.4. Zygomatic branches. The branching of the temporofacial division occurs within the parotid gland [9; 115]. The zygomatic branches run towards the lateral aspect of the orbit and remain inferior to the zygoma [9; 187]. The depth of zygomatic branches, 5 cm in front of their exit points from the parotid gland was found to be 9.6 ± 2.0 mm [452]. They supply the orbicularis oculi muscle (in common with temporal branches). The procerus, the zygomaticus major and minor, the levator labii superioris, the levator anguli oris, and the buccinator muscles can be innervated by buccal and/or zygomatic branches.

1.2.5.5. Temporal branches. The temporal branches are very superficial [452], especially when crossing the zygomatic arch. They are located between the temporoparietal fascia, which is the cephalic extension of the SMAS layer [503], and the deep temporalis fascia, the cephalic extension of SL-DCF [20; 503]. The exact location as the frontal branch crosses the zygomatic arch has been described with several lines and landmarks [9; 88; 187; 404; 421]. For Furnas the temporal branch crosses the zygoma within 5

38 mm of a line, joining the intertragic notch and the lateral extend of the eyebrow [187]. This corresponds to the crossing of a vertical line dropped from the anterior temporal hairline to the zygoma [187]. For Pitanguy and Ramos, the frontal branch can be located along a line drawn from the intertragic notch to a point located 1.5 cm from the lateral edge of the eyebrow [421]. For Castro Correia and Zani the frontal branch is within a triangular area, the summit of which is the earlobe and the sides formed by a line joining the lateral canthus and the lateral aspect of the "highest frontal crease" [88]. For Ammirati et al. [9] temporal branches cross the zygoma between 2.4 cm anterior to the tragus (1.5 – 3.5 cm) to 1.5 cm posterior to the lateral canthus. Probably the easiest landmark to remember is that of Ozersky et al. [404]: a line is drawn is from the orbital wall at the level of the lateral canthus to the tragus – the temporal branches are found 3 and 4 cm posterior to the orbit point. These last studies [9; 404] recognize that more than one temporal branch is present, a fact already shown by Davis et al. [115].

The depth of the temporal branches, 5 cm in front of their exit points from the parotid gland was found to be 2.3 ± 0.6 mm [452], confirming that these branches are the most superficial.

The temporal branches supply the anterior and superior auricular muscles, as well as the frontalis, the orbicularis oculi, and corrugator supercilii muscles.

39

Figure 9: Branching patterns of the extratemporal facial nerve The data were obtained from the dissection of 350 cervicofacial halves. The division patterns of the facial nerve are classified in types (I to VI) and the relative frequency of each type is marked. The description of each type is given in Table I and the text. Reprinted from Davis et al. [115] without permission.

40 1.3. Indications of parotidectomy The ideal procedure is to have a competent pathologist at one's elbow ready to advise and to make a frozen section diagnosis on a small piece of tumor tissue removed for biopsy Benedict and Meigs, 1930 [40]

The majority of parotidectomies are performed for parotid tumors, benign or malignant. The incidence of salivary gland tumors is about 5 for 100'000 people [158; 311]. Benign tumors represent 75% and malignant cancers 25% of salivary tumors [158; 311; 430]. The majority of tumors are located in the parotid gland – about 75% (Table II).

Accessory salivary Author Number of patients Parotid gland Submandibular gland glands

Eneroth [158] 2867 79% 7% 14%

Richardson [440] 752 82.5% 9.3% 8.2%

Batsakis [29; 30; 31; 32] 2640 73% 8.4% 18.6%

Spiro [488] 2807 70% 8% 22%

Table II: Distribution of salivary gland tumors (benign and malignant) in the three main salivary glands.

The distribution of benign and malignant tumors varies according to the site. An often quoted and easy to remember rule states that the proportion of malignant tumors in the parotid gland is 25%, 50 % in the submandibular gland, and 75% in accessory salivary glands. Table III gives the exact distribution in 3 large series.

41

Accessory salivary Author Number of patients Parotid gland Submandibular gland glands

Eneroth [158] 2867 18% 37% 47%

Richardson [440] 752 25% 26% 35%

Batsakis [29; 30; 31; 32] 2640 23% 44% 70%

Spiro [488] 2807 32% 55% 87%

Table III: Incidence of malignancy in salivary gland tumors in the three main salivary glands.

1.3.1. Histopathology of parotid lesions

To understand tumors, it is necessary to know all about them, to study large numbers brought together in groups, and to correlate the pathological findings with the clinical data, not only up to the time of operation, but thereafter so long as the patient lives. McFarland, 1936 [347]

The histologic classification of parotid gland remained controversial until recently, but now seems to be universally accepted, at least in broad lines [29; 30; 31; 32; 478; 515]. Table IV provides a list and the incidence of parotid gland tumors, as well as a brief description of treatment.

A non-tumor group of salivary gland pathologies is recognized and named tumor-like disorders. This group includes sialoadenosis, oncocytosis, benign lymphoepithelial lesion, cysts of the salivary glands, sialadenitis, and AIDS-related chronic lymphoid hyperplasia. If the correct diagnosis could be made with certainty, these lesions could be handled expectantly, if asymptomatic. Obviously, the frequency of these lesions in a given patient cohort will depend on the surgeon's attitude. An estimate can be obtained from Stoll et al. [500]: among 1009 parotidectomies, of which 841 were for benign lesions, the incidence of non-tumoral lesions was 21% of the total, and 25% of the benign lesions.

42

Foote Spiro Woods Spiro Kane NAME [178] [488] [558] [489] [267] TREATMENT Main All Parotid Parotid Locations considered salivary salivary Parotid (cancer) (cancer) glands glands Number of patients 776 2761 1360 380 194 1. ADENOMAS 66% 53% 84% Surgery is the main and only universally 1.1. Pleomorphic adenoma 58% 46% 60.5% accepted treatment. 1.2. Monomorphic adenoma 8% 6.8% 23% Universal agreement about the type of parotidectomy is still lacking; most often 1.2.1. Adenolymphoma (Whartin tumor) 6.5% 6.6% 22% superficial parotidectomy is the minimal 1.2.2. Others: operation for superficial 'lobe" tumors, while total - Myoepithelioma parotidectomy is used for deep "lobe" - Basal cell adenoma adenomas. - Oncocytoma Postoperative radiation has been debated in the past for pleomorphic adenoma, but has been - Canalicular adenoma abandoned in most centers. - Sebaceous adenoma Expectation is an option for poor surgical - Ductal papilloma candidates, and elderly patients. - Cystadenoma 2. CARCINOMAS 24% 47% 16% 100% 100%

2.1. Acinic cell 3% 3.0% 2.5% 14% 22% A A: Superficial or total Low grade: A; parotidectomy; preservation of 2.2. Mucoepidermoid carcinoma 12% 15.9% 47% 22% 4.5% High grade: B facial nerve; no irradiation. 2.3. Adenoid cystic carcinoma 2% 10.2% 2.1% 7% 15% B Treatment A should be changed to B, if tumor size > 4 2.4. Adenocarcinoma 4% 8.1% 3.9% 10% 13% cm (T3). - Polymorphous low grade B: Total parotidectomy; - Basal cell selective supraomohyoid neck dissection; resection of facial B - Papillary cystadenocarcinoma branches if grossly involved; - Mucinous postoperative irradiation. - Adenocarcinoma NOS (Not If tumor size > 6 cm : radical Otherwise Specified) parotidectomy with complete neck dissection. Carcinoma in pleomorphic 2.5. 5.8% 15% 7% B adenoma 6% 1.1% If extraparotid spread is present, involved structures 2.6. Squamous cell carcinoma 3% 1.9% 1.6% 6% 8% B should be resected. 2.7. Undifferentiated carcinoma 4% 1.3% 0.8% 1% 3% B 3. NON-EPITHELIAL TUMORS Angiomas, Lipomas, Neurogenic Surgery tumors, Mesenchymal tumors, Sarcomas 4. MALIGNANT LYMPHOMAS Radiation therapy ± Chemotherapy 5. SECONDARY TUMORS ?? 6. UNCLASSIFIED TUMORS ?? 7. TUMORS-LIKE LESIONS Sialoadenosis; Oncocytosis; Necrotizing sialometaplasia; Benign lymphoepithelial lesion; Salivary Variable indications for parotidectomy gland cysts (mucocele, salivary duct, lymphoepithelial, dysgenetic); Chronic sclerosing sialadenitis; Cystic lymphoid hyperplasia in AIDS

Table IV: Pathological classification of parotid tumors, their incidence (in percent) and brief description of their treatment The classification is modified from Sieffert et al. [478].

43 1.3.2. Preoperative work-up of parotid lesions

The majority of parotid tumors present as parotid mass [260; 261; 262; 267; 490], without other associated symptoms. The differential diagnosis between benign and malignant lesions cannot be made on clinical examination. The presence of associated pain (15% [490] to 21% [267]) or facial paralysis (12% [490] to 15% [267]) usually suggests a cancerous lesion.

Because of the risk of facial nerve paralysis, biopsy of a parotid lesion is usually condemned, although the exact risk has never been formally evaluated. Although fine needle aspiration of parotid lesions was described 60 years ago [499], its exact role is still debatable. Layfield et al. [306] reviewed studies published prior to 1986 and concluded that the false negative rate was 10% and the false positive rate 4%. Perioperative frozen sections are also subject to similar error rates, with false negative rates varying from 3% [267] to 16% [92].

The diagnostic performance of CT-scan or MRI has not been evaluated in a blinded fashion in a large series of patients [87]. For intraglandular masses, imaging can be helpful to assess the relationship of the tumor with the facial nerve. In large and extraglandular tumors, imaging can determine the extent of the parotid mass and show the involved structures, allowing for better patient information and surgery planning. MRI, because of its better assessment of soft tissue and multiplane images, is now the preferred imaging modality. If bone erosion is suspected, CT scan might better delineate the extent of bone erosion.

The diagnostic strategy should be individualized, depending on the resources available and the competence of the pathologist and/or radiologist. Without doubt, preoperative knowledge of the exact diagnosis of the parotid lesion will help the surgeon and patient to better plan the surgical procedure.

Practically, as stated in 1948 by Clausen and Henley [96], "with one exception any tumor of the parotid gland should be removed. When one considers the natural history of the tumor, it is apparent that should a tumor that grows very slowly appear in an individual whose life expectancy is short, one would be justified in observing its growth. In other cases, however, operation should not be delayed".

44 1.4. Surgical techniques of parotidectomy Severed facial nerves have recently been costing physicians $80,000 to $100,000 in judgments, even though no malpractice has been proven. The California courts have ruled that the patient's attorney can rely on the doctrine of res ipsa loquitur. Preservation of the facial nerve is, therefore, essential in benign parotid surgery. Too often, a surgeon is prone to excuse himself for a facial paralysis, on the basis that the nerve is too close to tumor tissue. In our experience, the consequences of a facial paralysis far exceed those of leaving behind residual benign tissue. Tabb et al., 1970 [508]

1.4.1. Parotid surgery mandates

The goals of parotid surgery are [232; 560]:

1) Prevention of recurrence, which requires a complete tumor removal, ideally with a cuff of normal parotid tissue, and without tumor seeding by spillage; 2) Facial nerve protection and preservation unless the nerve is directly involved by a neoplasm (almost always malignant); 3) Prevention of occurrence of Frey syndrome; 4) Prevention of other complications, such as salivary gland fistula, hematoma, wound infection, skin anesthesia; 5) Skin incision with optimal cosmetic results.

To achieve these goals the surgeon should have extensive knowledge of parotid pathology and of the exact clinical significance of each parotid tumor (see § 1.3.). Often the best aid for the parotid surgeon is the on-site availability of a competent pathologist with comprehensive knowledge, not only in parotid pathology, but also with thorough understanding of parotid surgery and the clinical implication of the various parotid neoplasms [40; 347].

A good parotid surgeon must have an excellent understanding of the complicated anatomy of the region (see § 1.2.) and in particular of the extratemporal facial nerve and its variations (see § 1.2.5.). The surgeon must know where the nerve is and where it is not [232].

Finally, the parotid surgeon should be an excellent "technician" [232], with enough skills and patience to dissect the tiny peripheral facial nerve branches, often displaced by the tumor.

45

1.4.2. Nomenclature of parotid operations

Traditionally, parotid surgery is divided in enucleation, superficial parotidectomy, total parotidectomy, and radical parotidectomy [408].

Enucleation (synonyms excision, extracapsular dissection, pericapsular dissection) consists in the resection of a parotid tumor without prior identification of the facial nerve and its branches. Essentially, the tumor is approached directly and resected with or without a cuff of normal parotid tissue. This was the original technique of removal of benign parotid tumors and has generated a large controversy (see History of parotid surgery; §1.1) because of numerous recurrences (see § 2.4.) in early publications [5; 40; 74; 281; 335; 344; 429; 494].

Superficial parotidectomy (synonym superficial lobectomy, lateral lobectomy) involves the ablation of the parotid tissue superficial to the facial nerve and its branches (the so-called "superficial parotid lobe"), after identification and dissection of the facial nerve and its branches. This approach, described by Barbat [24], Sistrunk [481], Adson and Ott [3], and later by Bailey [16] and others [79; 276; 335; 412; 436], is characterized by the identification of the facial nerve at the beginning of the operation. The branches of the facial nerve are afterwards dissected from the parotid tissue laying superficial to them. After this protection of the facial nerve branches, the suprafacial parotid [413], which contains the tumor, is resected. Usually, a more or less extensive cuff of normal parotid tissue surrounds the tumor.

Total parotidectomy starts with a superficial parotidectomy, followed by the dissection of the facial nerve from the underlying "deep parotid lobe". In this operation, the entire parotid gland is supposed to be removed, while the facial nerve is preserved.

In enucleation, superficial parotidectomy, and total parotidectomy, the anatomical and physiological integrity of the facial nerve is preserved. Sometimes, to underline this the unfortunate term conservative is added as in "conservative superficial parotidectomy"[408]. Although the term usually refers to facial nerve preservation, it has also been used to mean surgery less than superficial parotidectomy or enucleation [493]. A better but redundant phrasing is "with identification and preservation of the facial nerve". We think that these adjunctives should be abandoned and if cases where a branch of the facial nerve was sacrificed this should be specified.

Radical parotidectomy is a surgical operation on the parotid gland where the entire parotid gland and the facial nerve are removed en bloc with various amounts of surrounding, non-parotid

46 tissue. This operation, resulting in the sacrifice of the facial nerve, is usually performed for malignant tumors and is often associated with a reconstruction of the facial nerve by various grafting techniques.

While the terminology has been rather standard, unfortunately some different names have been employed in parotid surgery.

In 1950 Klopp and Winship used for the first time the term subtotal parotidectomy [276]. They realized that the parotid gland has numerous and variable extensions and "while these prolongations may frequently be removed in subtotal parotidectomy, the margin of excision is so slim that no claim should be made to total removal of all parotid cells" [276]. This statement is quite realistic and has been used by other authors [445; 568]. Unfortunately, the term subtotal parotidectomy has been used with a different meaning by Novotny and Pirozinsky [392] and more recently by Helmus [232]. Therefore, the term near-total parotidectomy, as proposed by Bron and O'Brien [61] seems preferable. In a near-total parotidectomy, a typical and complete superficial parotidectomy is performed and followed by a partial removal of the parotid tissue deep to the facial nerve. Obviously, the integrity of the facial nerve is preserved. The sometimes-used [392] synonym extended superficial parotidectomy should probably be abandoned. Also, the term selective deep lobe parotidectomy used by Leverstein et al. [315] seems similarly awkward.

In 1958, Patey proposed to name the divisions of the parotid gland suprafacial and subfacial parotid, and the operations suprafacial and subfacial parotidectomies [412]. Although this terminology is sound, its use was never generalized [168].

In 1982, Stevens and Hobsley [498] used several different terms. First, they employed the names primary and secondary parotidectomy for initial and revision parotid surgery. This terminology is acceptable and probably can be used. They also used the term semi-conservative parotidectomy for cases where part of the facial nerve is sacrificed during surgery. The use of the term "conservative" was discussed above and should be abandoned because of its ambiguity.

In 1984, Danovan and Conley [113] reviewed the weaknesses of traditional parotid operations: 1) the monobloc concept is violated in instances where the nerve enters the tumor or in deep lobe tumors; 2) often (60%) the neoplasm is in such proximity to the nerve that some form of limited capsular excision must be carried out, to preserve the nerve [113]. This statement and the results from other studies [214; 275; 299; 372; 411; 414] have encouraged surgery, less extensive than a complete superficial parotidectomy. Lyle [325], Vandenberg et al. [530], Yamashita et al.

47 [560], and Leverstein et al. [315] speak of "partial parotidectomy", O'Brien et al. [394] of "appropriate parotidectomy", Helmus [232] of "subtotal parotidectomy" for such a procedure. The use of this various terms and specifically subtotal parotidectomy should be condemned. Nevertheless, these recent publications of parotidectomies, which are not enucleations, but are less than a complete superficial parotidectomy, possibly correspond to a widespread clinical practice. Following Bron and O'Brien [61], the term limited superficial parotidectomy seems appropriate. In a limited superficial parotidectomy the facial nerve trunk is identified and the appropriate facial nerve, branches are resected before removing the tumor with a "sufficient" amount of normal parotid tissue.

Apparently, Bron and O'Brien believing that total parotidectomy is impossible without a radical approach, use the term total parotidectomy with the classical meaning of radical parotidectomy [61]. Again, this is unfortunate and the term should be used as classically employed and defined above.

Recently, Iizuka and Ishikawa [248] added a flurry of parotidectomy terms such as "extracapsular lumpectomy" which they define as “not essentially different from classical enucleation”. Why bother then to introduce a new and confusing term? They also speak of "segmental parotidectomy", of which they define four types, depending of the location of tumor relative to the main facial nerve branches. The exact purpose of such a subdivision of limited superficial parotidectomy is unclear, even in their article. Finally for them total parotidectomy should be called the "hamburger technique". Had they have cited Bailey [16; 17; 18] this naming would have had a historical interest.

In conclusion, the accepted parotidectomy terminology, namely enucleation, superficial parotidectomy, total parotidectomy, and radical parotidectomy, could be supplemented by two other terms: limited superficial parotidectomy, and near total parotidectomy. All other terms should be abandoned and journal editors should refuse their use unless a clear and heuristic purpose is explicitly defined.

1.4.3. General techniques of parotidectomy

1.4.3.1. Anesthesia Present parotid surgery is done under general anesthesia, without paralytic anesthetic agents. Curare derivatives are usually used for induction but their effects wear off within 20 minutes,

48 allowing for stimulation of the facial nerve and monitoring of facial movements or EMG monitoring of facial muscles.

1.4.3.2. Patient positioning The patient is positioned in a supine position, with the head and thorax elevated 10 to 20 degrees, in order to promote venous return from the head and neck and decrease vascular congestion. The head is rotated to the opposite side and slightly extended. If necessary, the table can be rotated 10-15 degrees towards the opposite side (Figure 10).

1.4.3.3. Infiltration We use local infiltration of an adrenaline solution (1:50'000) without local anesthetic [262; 492]. The solution is injected over the planed incision, in the subcutaneous dissection plane and vertically in the gland, around and away of the tumor. If the gland is not injected, solutions containing local anesthetics can be used [477] to optimize local vasoconstriction and decrease of systemic absorption of adrenaline.

1.4.3.4. Patient preparation Usually minimal retroauricular shaving is required. Ocular ointment is placed and the eyelids are taped. Skin disinfecting is accomplished with a solution that does not harm the cornea (for example: chlorhexidine 0.1%). After disinfecting, sterile electrodes for EMG based facial nerve monitoring are inserted in the orbicularis oculi and oris muscles. Reference electrodes are placed near the nasal alae.

1.4.3.5. Draping The entire half of the face needs to be exposed in the operative field in order to allow for the visualization of facial movements during surgery. Usually the drapes are placed close to the midline and a transparent sheet covers the eye and mouth. The entire neck needs to be exposed in the operating field for a possible neck dissection. A sterile cotton ball covered with an ointment is placed in the external ear canal to prevent the entry of blood during the procedure, since this can result in a postoperative external otitis.

1.4.3.6. Incision A review of the different incisions employed is discussed later (§ 2.3.1.). The usual incision has a sigmoid or a "lazy S" shape with a vertical preauricular limb, a curvature under the ear lobule, and a somewhat horizontal limb in a natural neck skin crease (Figure 11). The vertical limb follows established rules for scar camouflage by being placed at the junction between two facial esthetic

49 units - the pinna and the lateral face. A variation consists to hide part of this vertical limb in the concha, behind the tragus [37; 262]. The retroauricular portion of the incision should not be too narrow to prevent skin flap necrosis. The neck portion of the incision rarely needs to go beyond the level of the anterior border of the sternocleidomastoid muscle [520]. The exposure is adequate in most cases, the scar is small and well hidden, and the direction of the incision can be modified, should a neck dissection or extended base of skull resection [442] become necessary.

Recently, the use of a rhytidectomy incision with a horizontal limb near the neck hair line has been advocated and promoted as being more cosmetic [7; 11; 12; 100; 170; 210; 220; 235; 264; 512]. Whether this incision is worth the extra time and gives a more cosmetic result remains to be demonstrated. Obviously, modifications of the incision are required when skin needs to be resected because of invasion by a tumor, or previous parotid surgery [477].

1.4.3.7. The superficial skin flap While early parotidectomy techniques did not raise the superficial facial skin flap before beginning dissection of the gland [3; 24; 435; 481], it is now standard [477; 520]. The dissection of the flap is facilitated by prior infiltration. The flap can be raised in the subcutaneous fat (fat above and fat below), avoiding injuring the hair follicles. This plane of dissection is superficial to the SMAS and to the superficial cervical fascia (see § 1.2.1.). Dissection in this plane is advised if the tumor is very superficial and this plane of dissection is the safest with respect to the distal facial nerve branches. In addition, we and others [477] have the impression that dissection in the supra SMAS fat is easier, more expeditious, and results in less bleeding.

The dissection can also be made in a plane deep to the SMAS and thus to the superficial layer of the deep cervical fascia. While sub-SMAS dissection has been described as more esthetic and as a possible prevention of Frey syndrome (see § 2.2.8 and § 2.3.2.), the available data are not yet conclusive. An advantage of sub-SMAS dissection is the better vascularization of the skin flap. In sub-SMAS dissection, progression beyond the anterior extension of the parotid gland should be cautious, because here facial nerve branches become rapidly superficial.

The dissection can be carried out either with a scalpel [477] or with scissors [33; 556] with probably similar results. When using scissors, emphasis has been placed on doing the dissection parallel to facial nerve fibers and placing the scissors shafts at right angles to the underlying parotid gland [556].

50 Most of this dissection is "blind", since the skin flaps are put under tension in a posterior direction. The edge of the instrument can be seen as its tip elevates the skin. After elevation of the skin flap, stay sutures can be placed on the flap and the ear lobule to help the retraction [63].

1.4.3.8. Dissection of the posterior parotid The next step is to separate the posterior aspect of the parotid gland. The gland is easily separated from the sternocleidomastoid muscle, a step requiring sectioning the superficial layer of the deep cervical fascia [443]. At this point, it often becomes obvious that the anterior branch of the greater auricular nerve has to be cut. Some believe that the nerve should be tied to prevent the occurrence of a postoperative neuroma [477]. Deep to it is the digastric muscle, which is a key landmark for the operation as pointed out by Cadenat [80], and later by Janes [252], Redon [435], and others [335; 520]. The transverse process of the atlas is a useful palpation landmark for the identification of the digastric muscle [196]. The parotid gland needs to be completely freed from the digastric muscle, all the way to its bony insertion, which usually leads to the facial nerve trunk. Ligation of the carotid artery, recommended in earlier publications [16; 17; 58; 80; 149; 252; 312; 334; 341; 435; 492; 506], has been abandoned since the early 1950's [63; 272; 335]. In addition, the external jugular vein and the retromandibular vein should be preserved during the initial stages of dissection, in order to decrease the intraparotid venous pressure and thus decrease bleeding [443]. While freeing the digastric muscle, the subdigastric region should be examined for the presence of lymph nodes [477].

The gland is also separated from the anterior aspect of the cartilaginous external ear canal, a step that could result in the section of the auriculotemporal nerve, or its branches. During this step of the dissection the so-called cartilaginous pointer is identified, another landmark for the identification of the facial nerve. As already pointed out by Adson and Ott in 1923 [3], it is important to stay against the cartilage during this dissection, to decrease bleeding [196; 437; 477], to avoid injuring facial nerve branches within the parotid gland, and most importantly to use it as a guide during the dissection [108].

The dissection continues by the identification of the junction between the cartilaginous and bony external ear canal [443]. During this dissection, the posterior extension of the parotid fascia (see §1.2.3.) is sectioned, as it inserts on the tympanomastoid suture [196]. Redon had, in 1945, noticed this when he stated "A ce niveau, du tissu cellulaire condensé amarrant la glande au plan temporal protège à la fois et cache le facial" [435].

51 The tail of the parotid can be dissected before [556] of after [477] the preauricular part. In any case the dissection should not proceed too deep before the nerve trunk is identified. On the other hand, a common error is to start looking for the facial nerve too early [335]; the nerve is never superficial to the cartilaginous pointer, to the digastric muscle or to the mastoid tip.

1.4.4. Techniques for facial nerve identification

Before looking for the facial nerve, a good exposure and an excellent hemostasis is paramount. Good exposure is achieved by the previously described dissection of the posterior aspect of the parotid gland. The gland is retracted anteriorly by a retractor held by an assistant and the external ear canal is retracted backwards by a suture attached to a weight or to the drapes (Figure 13). Careful hemostasis is important because any blood in the wound will collect in its deepest portion, which is where the facial nerve should be searched (Figure 13). Above the level of the cartilaginous pointer, a regular monopolar electrocoagulator can be employed, while deep to this level a fine bipolar forceps is safer.

The key step in parotid surgery is the identification of the facial nerve (for facial nerve anatomy see §1.2.5.). In modern parotidectomy the nerve is identified at the trunk level and the branches followed forward in the glandular parenchyma [276; 335].

Useful landmarks for the facial nerve trunk include [477]:

1) the "cartilaginous pointer" or "tragal pointer", which is actually the anterior tip of the tragus portion of the external ear cartilage (Figure 13). The trunk is said to be 1 cm deep and 1 cm inferior to the pointer [262; 477]; 2) the posterior belly of the digastric muscle and its mastoid insertion, which is slightly lateral to the stylomastoid foramen; 3) the tympanomastoid suture, which can be appreciated by palpation [260; 262; 304; 394; 443]. The trunk is said to be 6 to 8 mm from the "inferomedial end of the suture" [237; 262; 362; 508], unfortunately, some controversy exists on what represents the end of the suture [443]; 4) the stylomastoid artery running with its vein a few millimeters lateral to the facial nerve. The stylomastoid artery arises from branches of the external carotid (see § 1.2.2) and follows the facial nerve in the stylomastoid foramen; 5) the styloid process which is located deep to the facial nerve [304; 443]. It can be palpated, but its visualization before identification of the nerve usually means that the nerve has been injured [237; 443].

52

Of these anatomical landmarks, the tympanomastoid suture is probably the most useful because it is a bony landmark and the least subject to variation [443]. The tympanomastoid suture is formed by edge of the tympanic bone lying on the mastoid bone. It starts at the posterior edge of the bony external auditory canal to extend on the undersurface of the temporal bone. It terminates by the stylomastoid foramen, through which the facial nerve exits the skull [443]. This relationship was first appreciated by Brintall et al. [59] and later advocated by others [237; 362; 443; 508]. As pointed out earlier, the parotid fascia inserts on the tympanomastoid suture. According to Robertson, the fissure can be followed safely until this fascial layer, which is often quite tough [362; 443]. The fascia needs to be divided close to the temporal bone and the dissection should proceed with gentle spreading in the direction of the facial nerve trunk, using fine mosquito forceps. The nerve usually appears as a white cord-like structure. Usually the stylomastoid artery needs to be coagulated with bipolar forceps prior to identifying the facial nerve trunk.

The nerve needs to be identified with certainty. One way to identify the facial nerve is to dissect the length of the trunk, until the bifurcation is clearly visualized. The length of the facial nerve trunk, i.e. the distance between the stylomastoid foramen and the division has been found by Dargent and Duroux as 13 mm [114] and 21 +/- 4.1 mm by Holt [239]. This dissection should be very cautious because accessory facial nerve trunks (actually a facial nerve that has bifurcated in the mastoid canal) have been described [187].

Another way is to use a facial nerve stimulator, as first mentioned by Clausen and Henley in 1948 [96]. In stimulating the facial nerve, it is important to use the lowest possible electric current, in order to avoid nerve damage and fatigue. The minimal amount of current necessary to elicit nerve stimulation and facial muscle contraction should be used. Bipolar stimulating electrode is more selective, uses less current for stimulation, and is therefore preferred. The use of "nerve pinching" as nerve stimulation technique [63] is probably to be proscribed, although a comparative trial of these two techniques has not been carried out.

Sistrunk [481] and Adson and Ott [3] proposed to identify the marginal mandibular branch and follow it to the facial nerve trunk. State [492] identified facial nerve branches in front of the parotid gland and used retrograde dissection to find the facial nerve trunk. McNealy and McCallister [353] also used the buccal branch, which was followed to identify the facial nerve trunk. Bailey [16; 17] and later Reissner [439] identified temporal facial nerve branches, as they cross the zygomatic arch. While the routine use of these methods has been abandoned, they remain useful in reoperations of the parotid gland where the scarring tissue renders identification of the facial nerve

53 trunk rather tedious. A few authors still favor the identification of a peripheral branch (usually the marginal mandibular nerve) and its dissection back to the facial nerve trunk [93; 440].

The branches the least difficult to identify are the temporal and the marginal mandibular, because they are often away from the field of dissection, and also because they exhibit the least variation.

The cervical branch can be found under the platysma muscle, at the lower border of the gland, between the sternocleidomastoid muscle and the angle of the mandible [93]. Usually it is identified by isolating the cervical extension of the retromandibular vein and dissecting it in a upward direction [93; 260; 303; 325]. However this relationship is inconstant in 10% of specimen, with the inferior facial nerve branches passing deep to the retromandibular vein [277; 297].

Finally, in difficult cases, the mastoid tip can be removed and the facial nerve identified in the descending mastoid segment [260; 492]. A more otologic approach was proposed by Heenemann [230], who advocated a retroauricular incision, with identification of the spine of Henle and dissection of the bony external auditory canal to find the tympanomastoid suture and stylomastoid foramen.

54

Figure 10: Positioning of the patient for parotid surgery

Figure 11: Typical parotidectomy incision

55

Figure 12: Dissection of the posterior portion of the parotid gland

Figure 13: Exposure and landmarks for identification of the facial nerve trunk

56 1.4.5. The superficial parotidectomy

Once the nerve is identified, the facial branches are dissected. The usual method is to dissect directly on top of a branch with fine hemostats or scissors (Figure 14). The spreading action is in the direction of the "faciovenous plane" [412], which is almost horizontal in supine position. During spreading, the superficial lobe parenchyma is then lifted and sectioned. This section should be performed with the dissected branch clearly in view. The instrument most often used is a number 12-scalpel blade, or more rarely scissors [556].

The section proceeds through the glandular parenchyma and usually results in some bleeding that needs to be controlled with a bipolar electric coagulator. A preferred option is to coagulate the parotid tissue prior to cutting. Fee and Handen [168] have advocated the use of the Shaw haemostatic scalpel, but a recent publication shows this device to be associated with higher postparotidectomy facial nerve paresis [426]. Others use the bipolar electrocoagulator [477]. We have found that bipolar scissors (Johnson & Johnson), which became recently available, have been extremely helpful in preventing this bleeding by providing vessel coagulation while cutting. Despite the close presence of nerve branches, no facial muscle movement, nor evoked EMG in the nerve monitor have been detected.

The dissection usually follows one branch and its tributaries to the anterior edge of parotid gland. Then the next branch is followed and so on. Some surgeons begin the dissection with the marginal mandibular and cervical branches and proceed in a superior direction [477; 556], others begin with the superior division, and some start at both ends to finish in the middle of the gland [520].

During this dissection, no traction on the facial nerve and its branches is applied. The advice of Woods [556] that "the gland may be avulsed from the nerve safely, because the gland will give before the nerve does", has not been widely accepted, probably because of fear to injure the facial nerve. For a superficial parotidectomy, the branches are left attached to the underlying parenchyma, thereby preserving their blood supply (Figure 15). Coagulation is performed with a bipolar coagulator, after reduction of the electric current. Usually, unless the bleeding is right on a nerve branch, no facial nerve lesions result from the bipolar coagulation. In cases of a bleeding very close to the nerve, a sponge soaked in epinephrine can be used.

57 Peripheral facial nerve branches become rapidly superficial at the anterior border of the gland. Special attention should also be paid in large tumors, which tend to displace the branches, making them more deep, more superficial, splaying them etc…

After the delivery of the superficial lobe specimen, it should be inspected for exposed tumor edges. While this cannot be always avoided, it might require an extra resection of deep lobe parenchyma, masseter muscle, or other adjacent non-essential tissues.

The handling of Stenson's duct is debatable: some ligate it [273; 492], others do not even look for it [520], and finally other surgeons think that it should be left in place to: a) drain the wound in the mouth [556]; b) preserve the function of the remaining deep lobe [196; 283]; c) prevent the development of postparotidectomy fistula [295].

58

Figure 14: Dissection of facial nerve branches during superficial parotidectomy

Figure 15: The operative field at the end of superficial parotidectomy

59 1.4.6. The total parotidectomy

In total parotidectomy, the facial nerve branches need to be dissected from the remaining deep lobe. Again fine scissors or hemostats are used to separate the nerve from the underlying parenchyma. The dissection is usually started with the lower, cervical, and marginal mandibular branches. The nerve branches should be handled as atraumatically as possible, for best post- operative results. We recommend against nerve lifting [435], but rather use depression of the parenchyma away from the nerve. If lifting is unavoidable, a metal hook held by the surgeon and only used for the actual dissection of the individual branch is recommended.

Each facial nerve branch, one after another, should be patiently dissected, before the deep lobe is free. Veins should be first ligated superiorly to avoid vascular congestion and decrease bleeding. The lobe extension behind the mandible and along the internal pterygoid muscle is thereafter fried, paying attention to the internal maxillary artery and pterygoid venous plexus.

1.4.7. Wound closure

At the end of the resection, the hemostasis is checked again and remaining bleeders coagulated, usually with a bipolar coagulating forceps.

Considering the nightmare of revision parotidectomy, it is useful to take a picture of the disposition of facial nerve branches prior to closure. This might also have some medico-legal value, demonstrating the anatomical integrity of the facial nerve.

The wound is then irrigated, preferably with warmed saline, a Frey syndrome prevention (see § 2.2.) barrier placed, the suction drain inserted, and the wound closed. If a SMAS flap has been developed it is carefully sutured at the posterior edge of the wound. Recently, we tend to routinely resect a portion of redundant skin from the anterior skin flap. The cosmetic value of this remains to be proven.

60 2. PAROTIDECTOMY COMPLICATIONS

Paralysis of the nerve produces disfigurement, with psychic trauma that many persons are unable to endure; in not a few instances, suicide has appeared to be the only escape. Beahrs and L'Esperance, 1956 [34]

The most important complication of any surgery for tumor removal is recurrence. The debate in the literature regarding the type of parotidectomy associated with the lowest rate of recurrence was eluded to earlier (see § 1.1) and remains somewhat debatable (see § 2.4.). The recurrence rates in various publications range from 0% to 70% [301].

Besides recurrence, parotidectomy complications can be subdivided in facial nerve paralysis (§ 2.1.), Frey syndrome (§ 2.2.), and a mixed group that we call wound complications (§ 2.3.). Wound complications include skin anesthesia, retromandibular depression, unfavorable and conspicuous skin incisions, and postoperative wound collections such as hematoma, seroma, and salivary fistula.

A list of publications describing the frequency of these complications is provided as Appendix 1. The great variation for the reported incidences of each individual complication can be easily seen. This does not even take into account the thoroughness with which complications are sought and the great variability of what each complication mean for different authors, as discussed below (see § 2.1. to § 2.4.).

61 2.1. Facial nerve

After any operation on the parotid there may be temporary paralysis. It is wise, therefore, to examine the face as soon as consciousness returns. Paralysis due to stretching is evident from the beginning and may require some months for recovery. Paralysis due to cutting appears at once and can only recover if sufficiently limited that intercommunication between branches of the nerve permits other fibers to take on the lost function. If only a segment of the nerve has been removed a nerve graft may be indicated. Janes RM, 1947 [253]

2.1.1. Historical background

Postoperative facial paralysis was once a necessary sequel of parotidectomy. Thomas Carwardine from Bristol was apparently the first to publish, in 1907, the case of a 18 year-old girl who had "a mixed tumor of the parotid which was shelled out so as not materially to damage the facial nerve" [84].

By patient dissection with small sharp scalpels the nerve was successfully freed and held up by a loop of catgut, then the vessels were tied above and below, and the whole gland was completely excised. One or two distal filaments were accidentally ruptured, but the greater part of the nerve was preserved. Following the operation there was a temporary facial paralysis, but in a couple of months the patient had twitchings at the angle of the mouth and kinesthetic sensations. Carwardine 1907 [84]

It is interesting to note that, although this case is usually credited to be the first case of facial nerve preservation, the patient had already parotid surgery (an enucleation) with nerve preservation. In addition, a temporary facial paralysis, which was almost routine until recently, resulted from "accidental rupture of distal filaments".

Publications on facial nerve preservation are sparse until the 2d World War. As early as 1914, Duval [149] performed the identification and preservation of the upper facial nerve branches in benign parotid tumors by removing the inferior aspect of the mastoid process and identifying the

62 nerve before its division. Barbat [24], followed by Sistrunk [481], Adson and Ott [3], and Saltzstein [455] identify the marginal mandibular branch of the facial nerve and follow it backwards to find the facial nerve trunk. Nevertheless, "although a temporary paralysis generally occurs from trauma to the nerve, it usually disappears within a year" [3]. It remained for Janes to describe the routine identification of the facial nerve trunk at the beginning of parotidectomy, which was followed by the dissection of the nerve branches forward [252]. With enucleation becoming out of favor, series of lack of mixed tumor recurrence and absence or low incidence of permanent facial paralysis appeared [35; 273; 527].

One of the first article to mention postoperative facial nerve results is Bailey, in his 1947 publication: "after operations on the parotid gland … a major degree of facial palsy always occurs; this may persist for weeks or months" [17]. The same year, Marshall and Miles reviewed the experience of the Lahey clinic and, already, find a higher incidence with malignant tumors (30% vs. 11%), with recurrent benign cases (22% vs. 8%) [334]. A role of the peripheral facial nerve anastomosis in the recovery of postparotidectomy facial nerve paralysis was recognized by Martin [335] and further elaborated on by Davis et al. [115]: because of paucity of anastomosis with other facial branches, the marginal mandibular branch is the most often injured branch. This has also been noted by others [168; 334; 336; 367; 394; 507]. Progressively, factors related to higher incidence of postparotidectomy paralysis were pinpointed [61; 336; 355; 367; 386; 394; 409; 411; 450; 507; 511; 529; 536; 538] (see § 2.1.5.).

As soon as postparotidectomy paresis was studied, some form of classification of the deficit became necessary and Patey and Moffat [411] classified the facial nerve deficits in 3 grades. "Patients were classified as Grade I if, on discharge of the hospital, either they had no facial paralysis or facial expression was so near normal that it was unlikely to give rise to comment during ordinary activities. We did not, however, regard weakness of depression of the angle of the mouth as by itself excluding a patient of Grade I, since the submandibular branch of the facial nerve is particularly liable to temporary interruption of its conductivity from any manipulation in the neighborhood. Grade II indicates that facial movements are present but were grossly and obviously weak. Grade III indicates complete or substantially complete paralysis." [411].

Looking at the, sometimes heated, controversies surrounding actual facial nerve grading systems (see § 2.1.3.) and the difficulties involved in establishing an objective facial function evaluation system (see § 2.1.2), it is easy to notice the evolution in the last 30 years.

63 One of the first, and almost unique, experimental study of factors involved in postparotidectomy facial nerve paralysis is due to Patey in the early 1960's [409; 411], so hints from other related fields needed to be reviewed (see § 2.1.4.).

While the present study was in progress, and facial nerve monitoring was routinely used [131; 133], two reports of the use of intraoperative facial nerve monitoring were published [399; 554] (see § 2.1.6.).

2.1.2. Techniques of evaluation of facial nerve function

Facial neuromuscular dysfunction is, stricto senso, an impairment of the function of the facial neuromuscular motor system. The deficits are complex but can be classified in to: 1) strength deficits (impaired motion of the facial muscles); 2) motor control problems (for example synkinesis); 3) relaxation difficulties (contracture and spasms); 4) psychological issues related to the inability to express emotional mimics [57].

Evaluation of the motor facial nerve function requires that movements of the facial musculature be elicited, either by a verbal command or by an external electrical stimulation. The appreciation of these movements represents the basis of every facial nerve evaluation system. This appreciation can be divided in subjective and objective (Table V).

The majority of external stimulation methods were developed in attempts to quantify the degree of Bell's palsy, early in its course, and to predict patients with unfavorable outcome. Subjective methods, such as nerve excitability threshold and the maximal stimulation test (Table V), have been superseded by objective methods, such as electroneuronography. Electroneuronography is seen as objective, because the response waveform can be stored and quantified [122], and therefore has been most widely used. However, recent work [101; 191] has revived interest in the older subjective electrical stimulation techniques. A detailed review of the electrical stimulation methods is beyond our scope, and can be found elsewhere [122; 191].

Electrical stimulation tests, notwithstanding their role in predicting the recovery of Bell's palsy, have definitive shortcomings when used in incomplete facial nerve paralysis. In facial paresis, electrical stimulation tests lack the necessary dynamic range for quantifying the remaining facial function. In addition, these test are concerned with the facial nerve in its entirety and have not been applied to evaluate the relative deficits of different facial neuromuscular territories such as

64 smiling vs. eye closure, for example. Finally, deficits other than motor strength are not addressed by these methods.

Therefore, an independent line of research has resulted in the development of various facial nerve grading systems [242; 243; 446], where facial movements are evoked by voluntary contracture. Although never spelled out, it seems that facial motor function is currently evaluated by two different systems, each addressing different extremities of the facial neuromuscular dysfunction scale [140]. The electrical stimulation tests are used for patients with little residual (0%) facial function, and voluntary evoked movements for patients with good residual (100%) facial function (Table V).

Stimulus Subjective Objective

Voluntary Facial nerve grading systems Topographic tests

External stimulation Nerve excitability threshold ENoG, Magnetic stimulation Maximum stimulation test

Table V: Classification of facial evaluation systems and their usefulness according to the degree of residual facial motor function From Dulguerov et al. [140] without permission.

When movements are evoked by voluntary contraction sources of error in the evaluation of facial function can be divided in those related to the production of the facial movements by voluntary contraction and those related to the evaluation of these movements. While evaluation problems have been subject to numerous studies [71; 72; 73; 171; 186; 243; 250; 356; 369; 377; 378; 416; 457; 555; 563], the production of facial movements has been assumed to be a reliable representation of facial nerve function. This supposes that: 1) the pertinent facial movements to be examined, really provide an adequate representation of the facial motor function; 2) the patient understands the required movements and can reproduce them reliably (negligible intra-subject variability); 3) variability of patient related factors is negligible across the patient population to be evaluated (negligible inter-subject variability); 4) the role of examiner in eliciting these movements is

65 negligible (negligible intra- and inter-observer variability) [140]. None of these assumptions has been formally tested.

In tests of facial neuromuscular function evoked by voluntary contraction the evaluation of facial movements can be classified in subjective and objective. Subjective evaluation methods correspond to the various grading systems discussed below (§ 2.1.4). The scoring in a the subjective facial nerve grading systems remains subjected to the variations of: 1) the production related variables discussed below; 2) the adequacy of a given grading system to apprehend the facial deficit; 3) the appropriate understanding and remembering (!) by the observer of the different grades that make up the grading system; 4) the observers' appreciation of the facial deficit; 5) the correct categorization of the deficit within a grading system; 6) the lack of observer bias [140]. In addition, the way the data are gathered by the observer (clinical examination, videotape, and photographs) could influence their assessment, as shown by Smith et al. [486]. To palliate to these inconveniences, objective methods were developed.

2.1.3. Topographic facial nerve function testing

Objective methods use some kind of measurement techniques, in the hope of reducing these errors and avoid biases in the evaluation. An ideal objective method, in order of importance, should: 1) not impede facial movements, therefore the face should not be touched during the movements; 2) be reproducible for a given individual, both in normal and pathologic cases; 3) provide synchronous data from the left and right side of the face, for comparison; 4) provide the measurements without touching the face; 5) provide absolute values (mm), not just percentages; 6) not require the observer to make the measurements, avoiding manipulation errors and bias; 7) be rapid; 8) well tolerated by the patients; 9) be stored in some form for later comparison, evaluation by other examiners, or further studies; and 10) not require markings on the face [140].

The methods used for objective topographic facial evaluation can be subdivided in three main groups: linear measurement, image subtraction, and miscellaneous techniques [140].

2.1.3.1. Burres' linear measurement studies In a pioneering article, published in 1985, Burres [72] assessed "seven standard facial expressions" using linear measurements in 20 normal subjects. Marks were placed on the face with a grease pencil, and measurements were taken on the patient with a hand caliper. The facial landmarks studied are shown in Figure 16 and the facial movements examined, with the measurements found to be relevant, are listed in Table VI.

66 This paper provides important information on the measurements to be included in the evaluation of specified facial movements (Table VI). Two movements were eliminated – frowning and soft eye closure. The measurements that were selected by the author for the evaluation of the different facial movements are [72; 73]:

1) Forehead wrinkle: SO – IO;

2) Tight eye closure: SO – IO and Na – IO (average of the two measurements);

3) Nose wrinkle: Na – L and Mc – L (average of the two measurements); ......

4) Smile: M – Mid and M – Ns (average of the two measurements);

5) Kiss: M – Lc.

A linear measurement index (LMi) is proposed [72] and elaborated on in subsequent publications [71]. This index was correlated with subjective facial nerve grading systems such as the one proposed by Fisch [174] and the first [242] version of the House-Brackmann scale (see § 2.1.3.3). Later, Croxson et al. [109] have demonstrated a correlation with the final version of the House-Brackmann scale [243]. Unfortunately in the calculation of this index, other measurements (corneal exposure, rest asymmetry) and somewhat involved ponderations were added. The rationale for these were never thoroughly discussed and this global index proposed by Burres has not gained popularity.

Nevertheless, the pioneer work of Burres has set the basic measurements to be looked for in an objective facial nerve evaluation system. In addition, these measurements were compared with the surface EMG, recorded over the cheek, lateral to the nasal alae. High correlation was obtained with closely related movements, such as smiling, nose wrinkling, and eye closure. Surface EMG has been shown for some time to be proportional to the force generated by the underlying muscles [45; 359] and the implications regarding the facial neuromuscular system have been discussed elsewhere [199].

In a subsequent publication [73], these linear measurements were applied to 44 patients with facial paralysis, the degree of which was unfortunately not specified, nor stratified. Nevertheless, differences in percent displacement between the normal and paralyzed side between 10 and 25% were found.

Overall, with some improvements and digital techniques, the linear measurements and even the index proposed by Burres could become a standard tool for the objective evaluation of facial

67 function. Unfortunately, the works that followed used different landmarks, facial movements, and evaluation techniques. Had they have provided a more elegant solution, progress would have been made.

68

Figure 16: Facial landmarks for linear measures Full circles indicate marks placed on the face (F, SO, IO, Na, L). Points F, SO, and IO are on a vertical line traced through the pupil. Point SO (Supra-Orbital) lies "on the most lateral portion of the orbital rim, above the pupil". Point F (Frontal) is 2 cm superior to point SO, and point IO (Infra-Orbital) is in "the most inferior fold of orbital skin, directly below the pupil". Na = nasion. L = lateral to the nasal alae. Natural landmarks are depicted with open triangles: Lc = lateral canthus, Mc = medial canthus, Ns = nasal spine, M = corner of mouth, Mid = midline of mouth. Modified from Burres [72].

69

Movement Landmarks Average Right – Left Average Percent Right – Left Coefficient distance at difference displacement displacement difference in of variance rest at rest displacement

SO - Mc 26 ± 5 mm 6% 34 mm 30 ± 11 % 9% 0.38 Forehead wrinkle SO - IO 37 ± 7 mm 3% 46 mm 24 ± 9 % 6% 0.38

Frown M - L 33 ± 9 mm 5% 37 mm 13 ± 10 % 7% 0.77

Eyes closure, soft SO - IO 37 ± 7 mm 3% 36 mm 3 ± 3 % 7% 1.11

SO - IO 37 ± 7 mm 3% 23 mm 38 ± 11 % 6% 0.18 Eyes closure, tight Na - IO 41 ± 8 mm 6% 31 mm 24 ± 7 % 6% 0.18

SO - IO 37 ± 7 mm 3% 27 mm 27 ± 12 % 7% 0.29

Nose wrinkle L – Mc 32 ± 7 mm 5% 24 mm 24 ± 8 % 5% 0.31

L – Na 47 ± 9 mm 4% 36 mm 23 ± 7 % 4% 0.45

Kiss M - Lc 72 ± 14 mm 3% 78 mm 8 ± 3 % 4% 0.34

M – Mid 28 ± 7 mm 3% 37 33 ± 16 % 5% 0.50 Smile M - Ns 39 ± 8 mm 5% 46 19 ± 10 % 6% 0.56

Table VI: Facial movements and the most meaningful measures taken (landmarks) The average distance at rest is the measurement at rest between the indicated landmarks (Figure 16), averaged across the subjects. The average displacement is the measurement after the specified movement between the indicated landmarks, averaged across the subjects. The percent displacement is the change in distance divided by the rest distance x 100. The coefficient of variance is the ratio of the standard deviation of the displacement to the mean of the displacement - it represents a comparison of the distance moved with the variation of this displacement, otherwise stated it is the signal-to-noise ratio in the facial displacement. A low coefficient of variance represents favorable landmarks. Selected data, showing the displacements with the lowest coefficient of variation, from table 1 and 2 from Burres [72].

70 2.1.3.2. Multi-camera linear measurement studies M. Frey et al. [186] used a sophisticated setup using 4 different cameras (Vicon, Oxford Metrics, United Kingdom). While the authors provide little technical details in their paper, the system was developed for the study of complex movements, is commercially available (100'000 SF) and quite sophisticated. The 4 cameras of the Vicon 370, probably used by the authors, are supposed to give the true three-dimensional evaluation of the movement by following 6 mm reflective marks glued on the skin. Such systems will probably become a reference for the evaluation and comparison of simpler facial function systems, which could be used, in routine clinical practice.

Frey et al. [186] studied markings (3 "fixed": tragus, chin, and central nose; 8 "dynamic": upper brow, upper and lower eyelid, nasal alae, philtrum, corner and upper and lower midlateral mouth) and movements (10) somewhat different from the ones used by Burres, without clearly justifying these choices.

Overall, the data shown and conclusions drawn are somewhat disappointing. The only interesting finding of their study was the presence of static points in the face during movements: the tragus (left and right), as well as a "point over the central nose". These points did not move more than one millimeter during the entire session – these were therefore used as reference points. It is, however, unclear what reference was used to determine movements of these points. Otherwise, the conclusions were similar to Burres' in that the points with maximal displacement were close to the area under study. While no rest distances or percent displacements are provided, a second important finding in this study is that data are provided to demonstrate little displacement of facial zones away from the area under movement.

In view of the cost of the Vicon system and the time involved in each facial measurement, the authors have developed a faciometer, which is essentially a caliper with a distance readout. No comparison between the data obtained with the faciometer and data from the Vicon measurements were provided.

2.1.3.3. Other linear measurement studies A) Fields and Peckitt [171] proposed to compare the distance between the lateral canthus and the corner of the mouth at rest and during smiling on both sides of the face. A "facial nerve

71 function index" (FNFI) is proposed as the ratio between the changes in distance between both sides:

Right Right ( Rest − DD Smile)

FNFI ≡ Left Left × 001 ()Rest − DD Smile

They did not state clearly how the measurement is to be performed and how much smiling is required and did not provide reference value for normal subjects or for patients with facial paralysis [171]. Finding that the FNFI had a skewed distribution, because of inherent intersubject facial asymmetry, additional calculations were added, in a subsequent publication [416], and called the facial nerve function coefficient. The new index had a more symmetrical distribution and a narrower range.

B) Murti et al. [369] proposed the so-called Nottingham system and compared it to the House-Brackmann grading system (see § 2.1.2.3.) and the index of Burres [72]. The Nottingham system is essentially a simplification of the linear measurements of Burres [71; 72; 73]: two movements (nose wrinkle and kiss) and several facial landmarks were eliminated leaving only four marks – SO, IO, L and M. A total score of this "linear measurement system" is a ratio of the sum of three distances (SO-IO for eyebrow rising and eye closure; and Lc-M for smiling) of each side.

To this mini-Burres system are added the letters Y or N depending of the presence of associated deficits such as hemifacial spasm, contracture, synkinesis, crocodile tears, decreased lacrimation, and dysgeusia. While such a melting pot of facial nerve function criteria might initially appear attractive, it is not obvious how crocodile tears etc. relate to the motor facial nerve function. In addition, the importance of these secondary deficits for the patient is unclear.

Beside such conceptual difficulties, there are numerous methodological shortcomings in this paper. First, while the title of the paper states that it is an "objective assessment of facial nerve function in the clinic", it is not clear how the proposed measurements are done. Second, while normative data could easily be obtained by looking at Burres' publication, the authors of the Nottingham system do not provide any for their new system. Third, no test-retest or inter- observer variability is shown and therefore the reliability of this system remains unknown. Fourth, the measurement of the distances on the face of the patient (as Burres was doing) is cumbersome and prone to subjectivity and observer biases.

72 C) In 1994, Johnson et al. [258] used a linear measurement technique derived from Burres' studies. Still photographs were obtained and 9 points centered around the eyes (F, Io, Na) and mouth (M, to which they added a philtrum and mentum point) were placed. Five movements were requested: brow lift, tight eye closure, smile, frown, and lip pucker. The measures were performed by projecting the slide on a digitizer board, a central coordinate obtained, and point coordinates obtained by pointing with a digital pen to the projected facial points.

A positive aspect allowing some normalization is the use of maximal movements that the patients were asked to sustain, i.e. they acronym MRSA for maximal static response assay. The authors obtained a standard deviation of dot positions on seven normal subjects of 0.07 cm. However, the magnitude of movements among subjects was widely variable, making it difficult to assign a normal range. The authors state that their assay is capable of differentiating expected movement from associated movements, photographs are reproducible when taken on different days, and their essay can differentiate between normal and paralyzed faces.

D) El-Naggar et al. [155] described the use of photographs printed on life-size black and white transparency films and state that photos are reproducible when taken on different days. No formal description of what is evaluated on these still images is given.

E) Isono et al. [250] used facial markings (10 per side + 4 midline) and one facial motion (eye closure). The session is taped and the tape digitized. The distance of displacement was measured using an image editing software. The reference point chosen was the nasal tip, which is probably not the best choice in view of the results of Frey et al. [186]. A sort of 2D x-y plots of the movements is given, but specific distances are not provided. It is not indicated if one of these distances is more relevant than the others, or if all need to be evaluated. Curiously, instead of using specific points of the upper head, all displacements on each side were summed and a side-to-side ratio obtained.

2.1.3.4. Image subtraction techniques A) Neely and coworkers have pioneered the use of digitized images for the evaluation of facial nerve function in 1992 [377]. The technique used is image subtraction: an initial image of the face at rest is stored and subtracted from subsequent videotape frames that are digitized. Probably, because of size of the files involved, gray scale images are used. According to the authors, "any area of the face that moves turns white and any of the face that has not moved zeroes out and turns

73 black" [378]. Software algorithms provide for defining areas on these facial images and counting "white" pixels within these areas. A "dynamic strength-duration curve" is generated.

Besides the use of digital technology, other improvements include the use of a head-holder to decrease head movements during the recording session and close attention to other recording parameters such as lightening, camera-head distance, and control of head rotation [377; 378]. The absence of such indispensable features in the majority of studies from other authors (see § 2.1.3.4.) renders their results almost meaningless.

In their initial study, large intersubject variation was found [377]. In their 1994 publication [378], computer modeling attributed the majority of the variability to intersubject differences, with little test-retest variability. While it is stated that the side of the face was also a variable, no data on this crucial point is provided.

In subsequent publications, Neely et al. demonstrated: 1) that facial displacements detected during movements do not follow an erratic coarse, but progress from a resting position to maximal contraction position [238]; 2) in the analysis of facial movements, their computer system is more sensitive that naive observers [238]; 3) the spatial resolution of their computer system is below 0.1 mm [231].

Apparently, this digital technique can be used for measuring facial movements. Nevertheless, because the system has received a US patent, the calculations are not discussed, and it is difficult to always determine what the authors are measuring. Finally, despite several publications it is unclear what are the important features that a computerized program should assess, in order to provide a useful tool for the analysis of facial function.

B) Sargent et al. [457] used a regular photo camera to obtain digital still images, which were than analyzed with a commercially available software (Photoshop, Adobe Systems Inc, Mountain View, CA, USA). Only 4 adhesive markings were used (SO, IO, M, Lc – according to Burres' naming) and only one facial movement (smiling) was used in normal controls, while 3 movements (smiling, eye closure, and forehead wrinkling) was used in paralyzed patients. The images were manually aligned based on the corneal light reflex and the distance obtained by getting the x and y coordinates, which were entered in a spreadsheet for calculation. An image subtraction technique of grayscale images was also used and the obtained surface manually outlined for area measurements. A ratio of areas to the normal side is provided as an index to correlate with the House-Brackmann (see § 2.1.4.) and the Nottingham scales.

74 While the methodology of this study has numerous shortcomings, mainly related to the number of manipulations required, with inherent sources of errors and observer subjectivity at each step, this technique could be improved and automated in the future. The authors conclude that the correlation between linear measurements and area measurements are not very good, and despite numerous shortcomings of the subtraction technique used, cast some doubts on this technique as claimed by Neely et al. Also, better correlation were found with the Nottingham scale than the House-Brackmann one (see § 2.1.4.).

C) Meier-Gallati [356] recently used a similar technique to that of Neeely et al., although they called it OSCAR, for objective scaling and area analysis. Three motions were used: smiling, eye closure, and forehead wrinkling. The face is divided in 4 zones (2 for the lower face) on each side. Movements (both voluntary and synkinetic) were calculated as changes in the light reflection patterns within each zone. An overall index is provided with the smile and eye closure counting as 40% each, and the forehead wrinkling as 20%.

Scores of about 95% were found in 12 normals, while patients with complete facial paralysis had a score around 5% for smiling and forehead wrinkling. Eye closure was problematic in paralyzed patients and was therefore excluded from the computation of the index! A comparison with HB scores was given in 12 other patients with facial paralysis.

Drawbacks include the necessity of ambient luminosity control, fixed subject-camera distance, and long duration of the procedure (10 minutes). Probably the greatest problem with this method is necessity to keep the head almost absolutely still, in order to avoid image subtraction measurements being erratic. To prevent this a special head-containing device needed to be built to prevent head movements. Still, it seems quite difficult to request a patient to remain for 10 minutes in a head restrain.

In conclusion, the image subtraction method is technically difficult, and the interpretation of the available results difficult to evaluate. The data from Neely et al. probably attest that the measurements are technically possible. One obvious advantage is the use of area measurements rather than distance measurements as proposed by Burres et al. Until the advent of a commercial "package" and a blinded comparison of the two techniques, the controversy will continue.

75 2.1.3.5. Miscellaneous techniques A) Oshyama et al. [398] used 19 surface electrodes to record simultaneously the EMG over the entire face. An overall score is obtained using an "interpolation formula" and scores on each side of the face were compared. They correlated these EMG scores with scores according to Yanigahara's classification. The technique is interesting but was never popularized, probably because it is quite cumbersome.

B) Wood et al. [555] evaluated two facial movements, namely eyebrow elevation and smiling, in 11 normal subjects using microscaling. Microscaling is an analog technique allowing the placement of lines on a video screen and the measure of the distance between these lines is provided. Practically the videotape is stopped on an image at rest, a first line is positioned, the tape advanced to the maximum movement frame and a second line placed, providing the desired lines. The distance between these lines is than measured.

They provide data on test-retest on the same day, test-retest on different days, side-to-side and intersubject variations (Table VII). Notwithstanding, the shortcomings of the method, the intrasubject variability appears to be around 5%, while the intersubject variability is high, close to 25%.

Shortcomings of this method include: 1) the requirement that the head stays in stable position relative to the camera which is almost impossible because people tend to move, rotate, elevate the head, even when required to stay still; 2) the element of subjectivity introduced by the observer required to judge the exact position of the facial contour that is evaluated. This makes measurements around the nasal alae or in midface almost impossible.

This is one of the first studies to assess test-retest variability in the evaluation of facial function. However, and because of the shortcomings of the method, we disagree with the authors' conclusion that in view of the large intersubject variability, " using objective measurements alone to follow facial palsy over time may not be valid".

76

Movement Factor Average variability [%] Range [%]

Test-retest same day 4.06 1.39 – 13.08

Test-retest different day 5.18 1.16 – 26.66 Brow Side-to-side 6.04 0.00 – 17.43

Intersubject 25.26

Test-retest same day 5.09 2.79 – 8.81

Test-retest different day 6.00 2.74 – 13.72 Smile Side-to-side 13.17 0.00 – 48.75

Intersubject 22.88

Table VII: Variability of mouth and eye movements in 11 normal subjects using microscaling Modified from Wood DA et al. [555]

C) Yuen et al. [563] applied moiré topography to the analysis of facial movements. The technique is not new and requires a special camera, but has the advantage of being a non-touch technique: the face is filmed and no markings are necessary. Apparently, Yuen et al. [563] used still images, which were printed for analysis. In a way, the technique remains somewhat subjective because moiré lines are counted by the observer. The authors defined "indexes" around the inner canthus, corner of mouth, and nasolabial groove. These indexes are essentially ratio of the number of lines on the right and left sides. Apparently, these indexes are different in healthy versus paralyzed individuals, however no cutoff limits are specified and the technique is not compared to other evaluation methods of facial function.

While the non-touch aspect of the moiré technique is interesting and future computerized systems (the study of Yuen et al. [563] is not) might be able to provide direct counts of the number of lines, several shortcomings are inherent of the technique. Moiré lines are provided according to distances from the camera and, therefore, what can be counted is changes in this distance. How these changes in the anteroposterior dimension are related to facial movements, the majority of which are in the frontal plane, remains to be determined.

77 2.1.3.6. Conclusions It is difficult to draw firm conclusions from these diverse studies. The topic is rather new, with almost all studies been published during the 90's, while our study was in progress. Probably, the only firm conclusion is that there seems to be an increasing use of digital imaging techniques. Most probably, with the recent advent of commercial digital video recorders, any future system that might get wide acceptance will be digital and use computer software to measure and calculate some form of facial function index [140].

Two major questions remain unanswered: 1) what should we be measuring? 2) how should the measures be done?

78 2.1.4. Facial nerve paralysis grading classifications

Facial nerve grading systems can be divided into gross and regional. In gross facial nerve grading, the overall facial function is evaluated, in a sort of synthetic gestalt. In regional grading systems, different areas of facial function are examined and scored separately. In some regional grading systems the different area scores are summed to provide an overall total score. Regional systems may be weighted or unweighted. In weighted regional grading systems, the scores of some facial areas are multiplied by a higher coefficient, because these areas (usually the eye and mouth) are though to be more important that other facial regions.

In 1983 John House, in his "Triological" thesis [242] reviewed 8 published facial nerve paralysis classification systems, including gross systems proposed by Botman and Jongkees [55] (Table VIII), May [242; 339], Peitersen [418] (Table IX), the regional systems of Smith [242], Adour and Swanson [2] (Table X), Janssen [242], Yanagihara [561] (Table XI), and Stennert (Table XII). While the commentaries by the author, on each individual scale were rather subjective, the interest of House's work [242] was the grading of facial movement using all 8 scoring systems, by 15 facial nerve experts. The patient facial movements were videotaped and identical tapes were mailed to the experts who scored each tape according to the 8 scoring systems.

The following conclusions were drawn: 1) good overall agreement was present between scales, with high concordance coefficients and high correlation; 2) the variation in scoring was low for normal facial function and for severely paralyzed patients, the major was for intermediate facial deficits; 3) regional scales show higher reliability than gross grading scales; 4) "the facial experts" preferred to use gross facial nerve evaluation scales; 5) facial experts felt that secondary deficits (synkinesis, facial muscle spasms, tics and twichings, "crocodile" (gustatory) tears, taste and tearing disturbances) should be included in the facial nerve grading scales.

Because of the conclusion that gross scales are easier to use, House [242] proposed a new facial nerve classification system. The grading system was later modified [243] and called the House-Brackmann grading system (HB). Since of its endorsement by the Facial Nerve Disorders Committee of the American Academy of Otolaryngology – Head and Neck Surgery, this system has enjoyed a large popularity and has become sort of a standard system for the evaluation of facial nerve paralysis (Table XIII).

While standards are useful by allowing for comparison across studies, several studies cast doubts on the superiority of the House-Brackmann scale [369]. Smith et al. [485; 486] compared

79 the inter-observer agreement with the same grading systems: while the overall inter-observer agreement was found to be reasonably good (κ: 0.5 to 0.65), no single system was clearly superior to another [485], and gross grading systems had somewhat lower agreement scores [485]. Croxson et al. [109] compared the House-Brackmann scale with the Burres' LMi and found a good correlation.

Another grading system was recently proposed by Ross et al. [446] (Figure 17). Contrary to the House-Brackmann scale, which has only 6 choices, this scale has a maximal score of 100 and a minimal of 0 (sort of percentage). The asymmetry at rest and synkinesis are subtracted from the voluntary score. This grading system was correlated with the House-Brackmann scale and with improvements of the patient's paralysis.

In the present era of quality of life studies, it remained to propose an evaluation of the disability and problems, as seen by the patient. Van Swearingen and Brach [531] filled the gap by proposing a self-rating questionnaire where the questions are directed to the patient. This questionnaire addresses not only the physical disability but also the emotional and social aspects.

Botman and Jongkees [55] classification

0 Normal Normal facial muscle activity

1 Light paresis Normal at rest, during speech. Eye closure possible. Asymmetry of the lips during laughing and whistling

2 Moderate paralysis Normal at rest. Eye closure impossible. Asymmetry during laughing and talking.

3 Severe paralysis Asymmetry at rest, dysfunction during movement.

4 Total paralysis No tone at rest. Total loss of function.

Table VIII: Botman and Jongkees – a gross facial nerve paralysis classification.

80

Peitersen [418] facial nerve paralysis grading classification

0 Normal No associated movements

1 Light paresis Slight palsy contracture less than 1 mm, i.e. just visible without associated movements

2 Moderate paralysis Clearly visible contracture and associated movements

3 Severe paralysis Disfiguring contracture and associated movements.

4 Total paralysis Atonic facial paralysis without contracture or associated movements.

Table IX: Peitersen – a gross facial nerve paralysis grading classification.

Adour and Swanson [2] classification

Percent of functional deficit SITE 0 0 – 25 % 25 - 50 % 50 – 75 % 75 – 100 %

Forehead 0 +1 +1 +2 +2

Eye 0 +1 +2 +3 +4

Mouth 0 +1 +2 +3 +4

Table X: Adour and Swanson – a regional weighted facial paralysis classification. A precise point relating to two muscular groups is determined. The distance that the forehead or mouth can be moved are measured and assigned values within the four quartiles. The degree of eye closure is estimated.

81

Yanagihara [561] classification - 5 or 3 points

At rest

Forehead wrinkle

Blink

Eye closure: light For every movement, a score is attributed. In the 3 points scale the scores range from 0 to 2, while Eye closure: tight in the 5 points scale the scores range from 0 to 4. The lowest score is 0. Eye closure: involved side A global score is obtained by summing the Nose wrinkle individual scores.

Whistle

Grin

Lower lip depression

Table XI: Yanagihara – a regional weighted facial paralysis classification. This scale is a weighted one because the eye closure function is evaluated 4 times.

82

Stennert [495] classification

Resting tone Difference in palpebral fissures > 3 mm Ectropion Loss of nasolabial sulcus Drop of the corner of the mouth

Motility – upper face Frowning less than 50% compared to the normal side Incomplete lid closure: slight innervation (as during sleep) Incomplete lid closure: maximal innervation

Motility – lower face Canine teeth not visible Second upper incisor not fully visible Distance between philtrum and corner of mouth during whistling: decrease of less than 50%, as compared to the normal side

Secondary defects Hyperacusis: present Impaired gustation Synkinesis between more than 3 areas (forehead, eye, nasolabial sulcus, corner of mouth, chin) Spasm: present Spasm: strongly present Spasm: inconvenient Lacrimation: less than 30% of normal side Lacrimation: absent Contractures: present Crocodile tears: present

Table XII: Stennert – a regional facial nerve paralysis classification. A facial paralysis score is calculated by adding the number of positive signs in resting tone and motility of the upper and lower face; this number is multiplied than by 10. The secondary defect facial paralysis score is calculated by adding the number of secondary defect signs and multiplying this number by 10.

83

House-Brackmann [243] classification

I Normal Normal facial function in all areas

II Mild dysfunction Gross: slight weakness noticeable on close inspection; slight synkinesis possible At rest: normal symmetry and tone Motion forehead: moderate to good function Motion eye: complete closure with minimum effort Motion mouth: slight asymmetry

III Moderate Gross: obvious but not disfiguring asymmetry; noticeable but not severe synkinesis, dysfunction contracture and/ or facial spasm At rest: normal symmetry and tone Motion forehead: slight to moderate movement Motion eye: complete closure with effort Motion mouth: slight weakness with maximum effort

IV Moderately severe Gross: obvious weakness and/or disfiguring asymmetry dysfunction At rest: normal symmetry and tone Motion forehead: no motion Motion eye: incomplete complete closure Motion mouth: asymmetric with maximum effort

V Severe Gross: only barely perceptible motion dysfunction At rest: asymmetry Motion forehead: no motion Motion eye: incomplete closure Motion mouth: slight movement

VI Total paralysis No movement

Table XIII: House-Brackmann – a gross facial nerve paralysis classification.

84

Figure 17: Ross – Nedzelski – the latest described gross facial nerve paralysis classification. From Ross et al. [446] without permission.

85 2.1.5. Physiopathology of facial nerve paralysis

Iatrogenic nerve injuries undoubtedly have occurred since antiquity…Nonetheless, little information is accessible concerning their (1) overall incidence among peripheral nervous systems lesions and (2) relative incidence among the general category of iatrogenic disorders. Wilbourn, 1998 [548]

Paralysis after parotidectomy could results from: 1) deliberate section (sacrifice) of the facial nerve or its branches; 2) inadvertent but recognized section of the facial nerve or its branches; 3) unclear causes while the anatomical integrity of the facial nerve is intact [409].

The reasons for sacrificing the facial nerve or its branches should not be discussed here (see § 1.3. for some hints). The inadvertent section of the main facial nerve truck or its branches is possibly related to the competence of the surgeon and its reporting to his honesty to recognize it.

The most frequent causes for postparotidectomy paralysis belong to the third group. The different factors that could be evoked include: 1) ischemia; 2) mechanical trauma, such as pressure, stretching, or crush injury by instruments; 3) heat damage from the use of an electrocoagulator; 4) cooling injury; and 5) nerve toxic substances [168; 401; 409; 411].

While some of factors involved have been studied in experimental animals, in the context of the physiopathology of nerve injury, little has been done for surgical operations, and even less specifically for parotidectomy. The relevant anatomy of a peripheral nerve and its microvasculature is depicted in Figure 18.

2.1.5.1. Ischemia Peripheral and cranial nerves have a dual blood supply: an intrinsic system consisting of longitudinal small vessels within the endoneurium and an extrinsic system of regional arterioles and venules entering the epineurium. The anastomosis of both systems forms a rich intraneural microvascular network, present in all neural layers [324]. This extensive vascular network provides peripheral nerves with a generous blood supply. In addition, the metabolic demands of peripheral nerves (0.01 cm3 O2/cm3/tissue/min) are much less than those of fully saturated arterial blood (0.03 cm3 O2/cm3/tissue/min) [90]. Furthermore, supply of oxygen and other nutrients by diffusion from surrounding tissue could supplement significantly the nerve in ischemia. Finally,

86 peripheral nerve fibers can function for considerable periods of time using anaerobic metabolism. Because of these factors, ischemic peripheral nerve damage has been difficult to accomplish even experimentally [90].

Figure 18: Microanatomy of peripheral nerves and their vascular supply. The Schwann cells enveloping nerves fibers are covered by a fine membrane, the endoneurium (end). Bundles of nerve fibers are further collected in fascicles, surrounded by perineurium (p), which is a mechanically strong membrane. Fascicles are embedded in various amounts of loose connective tissue – the epineurium (epi). The extrinsic vessels (exv) support regional vessels (rv) which penetrate the epineurium and branch into a longitudinal vessels running in the epineurium. Epineural vessels pierce the perineurium in an oblique fashion (open arrowhead) to feed vascular plexus in the perineurium. Within the endoneurium, the vessels are mainly capillaries. The intrafascicular space has a constant biochemical composition, maintained by the perineurium cells and endothelial cells of the capillaries of the endoneurium, which can be regarded as constituting a blood-nerve barrier. From Lundborg et al. [324] without permission.

87 For example, to produce ischemic damages of the cat sciatic nerve, all segmental arteries needed to be ligated and the nerve cut on both sides of the area under study [42]. Conduction failure appeared in 30 minutes and was hastened by wrapping the nerve [42]. For rat sciatic nerve, Schmelzer et al. [466] needed to ligate the iliolumbar, inferior mesenteric, and both iliac arteries, as well as the abdominal aorta. Even with such an extensive devascularization, compound action potentials returned to baseline after one hour of limb anoxia (metabolic conduction bloc) [466]. After 3 hours of anoxia, the nerve compound action potentials, did not regain normal values, even after a follow-up of one month [466].

Chalk and Dyck [90] summarized the experimental findings regarding peripheral nerve ischemia: 1) peripheral nerves are quite resistant to ischemic damage, requiring extensive impairment of the nerve blood supply, 2) conduction failure occurs after 30 minutes of ischemia, 3) nerve fibers located within the center of the nerve are more vulnerable that peripherally located neurons, 4) nerve trunks exhibit regional variations in ischemia susceptibility, probably related to their segmental blood supply, 5) the axon appears to be the most vulnerable structure, 6) demyelination is secondary to axonal injury, 7) the deleterious effects on axons appears to result from the interruption of the fast axonal transport mechanism.

From these experimental data, it seems highly unlikely that postoperative nerve paralysis in routine surgical operations ever results from nerve ischemia, secondary to nerve devascularization.

2.1.5.2. Mechanical trauma – compression injuries An excellent review the intraneural events associated with compression can be found in Lundborg et al. [324]. At compression pressures of 20 –30 mmHg, an impairment of the blood flow in epineural venules is noted, which leads to an injury of the endothelial cells of endoneural capillaries. The resulting increased capillary permeability in turn leads to leakage of fluid and proteins into the intrafascicular space, producing edema and an increased intrafascicular pressure. The lack of lymphatic vessels within the intrafascicular space and the presence of a blood-nerve barrier render the absorption of this edema difficult [324].

When compression at pressures of 30 – 80 mmHg is maintained for 2-4 hours, endoneural pressures increase to 3 times the baseline have been measured. This increased endoneural pressure results in an occlusion of the oblique perineural vessels (Figure 18), further decreasing the blood flow to axons and producing a localized hypoxia [324]. Pressures of 80 mmHg have been found to result in complete blood flow occlusion [397].

88 The presence of endoneural edema has been associated with electrophysiological functional dysfunction. Conduction is blocked, because of a localized anoxia of the axons. At these pressures, axonal transport is blocked, resulting in a depletion of essential substances within the distal segment of the neuron. When the compression remains at these rather low levels of pressure and if the duration of compression does not exceed few hours, the nerve function recovers relatively rapidly (metabolic conduction bloc).

When high compression levels are applied to peripheral nerves, the above-described local ischemic events are associated with mechanical deformation of the nerve trunk itself. The main event here is a focal demyelination [123] in the area under compression resulting in pushed edges of myelin sheets [395], while axons remain intact. Larger fibers seem to the first to be demyelinized [395]. The recovery with this "demyelination conduction block" is prolonged for 6-12 weeks, corresponding to the recovery time frame of Seddon's neurapraxia [472] and Sunderland's first degree of nerve injury [505])

2.1.5.3. Mechanical trauma –crushing The acute nerve crushing injury, such as squeezing with a smooth tipped forceps, leads to axonal interruption with preservation of the continuity of the nerve. The fate of Schwann cells is somewhat variable, but, while the myelin sheets are pushed to the edges, in most cases some form of continuity is maintained. This corresponds to Seddon's axonotmesis [472] and Sunderland's second-degree injury [505]. The recovery takes place by the regeneration of the severed axons across the site of injury. If the Schwann cells have remained intact, the recovery is usually satisfactory since regenerating axons regain their initial pathway and are guided to their initial targets [65].

Severe and prolonged compression or crushing injuries can result in a loss of continuity of the connective tissue envelopes of the nerve trunk, such as the endoneurium (Sunderland's third degree of nerve injury [505]), the perineurium (Sunderland's forth degree), and the epineurium (Sunderland's fifth degree). These correspond to neurotmesis in Seddon's classification [472].

While nerve crushing has been extensively used for studying nerve trauma and regeneration, the exact differences with the nerve compression model are unclear and little was done to standardize the mechanism for generating the nerve trauma. Recently, the force produced by a Halsted mosquito clamp was measured as varying between 17 N for the first clamping notch and 40 N for the third one [374]. Similarly, the force generated between the teeth of an Adson forceps

89 was 20 N [374]. Application of such compression forces by crushing the facial nerve for 10 minutes reliably produced axonotmesis [374]. If the nerve surface crushed is estimated to be 1 mm2, the corresponding pressures are in the order of 20 to 40 106 Pa (or MPa). The exact effects of shorter-lasting crushing injuries, while not studied extensively, are probably similar.

14

C 12

10

8

6

Stress [MPa] Stress 4

2 B D E A 0 0 102030405060

-2 Strain [%]

Figure 19: Stain-stress of peripheral nerves. Typical stress (Force/Area) to strain (elongation) curve of peripheral nerves (modified from Haftek [219] and Grewal et al. [208]).

2.1.5.4. Mechanical trauma –stretching Peripheral nerves have been found to follow a peculiar stress-strain curve (Figure 19) [219]. It is common knowledge, that when nerves are cut, their edges retract. This indicates that nerves present some natural stretching, that has been called in-situ stress (Figure 19 A) [208]. Measurements on rat tibial nerves have shown that this corresponds to a strain (degree of elongation) of 11% and a stress (force/area) of 0.05 ± 0.03 106 Pa [208]. Also, at rest, the nerve fibers follow a tortuous course within the fascicles, forming dark bands, called bands of Fontana, in light microscopy. When the nerve is elongated, the fibers straighten and some epineurium elongation takes place (Figure 19 A to B) [219]. Again for the rat tibial nerve this corresponded to a strain of 20% [208]. If further stress is applied, the nerve behaves as an elastic material (straight line – Figure 19 B to C) until the elastic limit is reached (Figure 19 C). The elastic limit for rabbit tibial

90 nerve was found to be at a strain of 38% and a stress of 11.7 ± 0.7 106 Pa [208]. This corresponds to loads (forces) of 5 N or 500 g [219]. After the elastic limit, the nerve integrity is lost and nerve structures rupture progressively, at the elastic limit, the epineurium [219] or perineurium [208], followed by the axons, myelin sheets, and endoneurium. Finally, the entire nerve is severed (Figure 19 E).

Obviously some stretching is necessary to avoid nerve damage during normal limb joint motions. Initially it was thought that up to the elastic limit, the nerve was able to regain its original length and no functional impairment issued. However, elongations of only 6% have been found to produce longitudinal tears of the perineurium, with nerve fascicles bulging through (see Grewal et al. [208] for a review). Despite this, the nerve appears grossly normal, to a naked eye! The perineurium is the main component of the nerve-blood barrier and is responsible for the maintenance of constant intrafascicular environment. Rupture of the perineurium will result in leakage of proteins in the fascicles and intrafascicular edema. The exact significance of these changes and the elongation limit beyond which irreversible functional changes occur has yet to be determined precisely [208], but is probably way below the 38% elongation that characterizes the elastic limit.

Beside these mechanical lesions produced by stretching, a 15% elongation of rat sciatic nerves resulted in complete occlusion of the nerve microcirculation [324]. If stretching was maintained for less than 30 minutes recovery was possible, while longer periods resulted in definitive ischemic injuries. In addition, similar amounts of elongation result in rapid electrophysiological dysfunctions: the compound nerve action potential is decreased and never recovers completely (see Grewal et al. [208] for a review).

To summarize these findings from animal data, it is reasonable to conclude that the elongation limit above which stretching results in irreversible nerve damage is around 10 to 15%. This limit is probably only valid for short lasting stretching, with prolonged elongation generating damage at lower stresses.

2.1.5.5. Cold injury Schaumburg et al. [463] cooled cat sciatic nerve to 20 °C and this did not produce any clinical, electrophysiological or effect. At 10 °C, some nerve damage was produced, but clinical recuperation was apparent after 2 days. At temperature below 7 °C, the damage was permanent. Duration of exposure (0.5 – 2 hours) had little effect on the results. Franz and Iggo [181] found that conduction decreased linearly by about 3% by degree Celsius below 17°C. Histologically,

91 edema, as well as abnormalities of endoneurium capillaries, Schwann cells, and axons were noted, resulting in axonal degeneration and demyelination [26]. A breakdown of the blood-nerve barrier was also present [393], but the precise physiopathological sequence of events is still undetermined.

While data on peripheral nerve injury from an actual surgical procedure is rarely available, cold injury to phrenic nerve during cardiac bypass surgery was studied by Rousou et al. [448]. Exposure to ice slush or a cooling jacket with outside temperature of 5°C resulted in an incidence of phrenic nerve paralysis of 24%. Recovery took place in 2-3 months. It is unlikely, however, that surface temperature of routine surgical procedures reaches these low levels.

2.1.5.6. Damage from electrocautery While electrocauterization has been used for close to 100 years, relatively little is written on possible nerve damages associated with their use. The electrical current delivered by electrocautery units flows from the active to the return electrode. In monopolar units, the return electrode is remote and large, resulting in low local current densities, while because the active electrode is small the local current densities are large. In bipolar systems the return electrode is close to the active one and, because of the limited current spread, low currents are required for hemostasis. As current flow through tissue, the electrical impedance to current flow results in to the transformation of electrical energy into thermal energy (Joule's law) [453]. This thermal energy produces various effects such as protein denaturation (60 °C), cell water evaporation (>100 °C) resulting in either cell explosion and vaporization or cell desiccation, depending on the waveform and energy of the delivered current [453]. Unfortunately, the power delivered by electrocautery units is poorly standardized.

Physiopathologically, electrically induced neural damage has been classified into: 1) thermal damage due to heat production, 2) transient nerve dysfunction secondary to the passage of current through nerves [153], 3) delayed damage from vascular injury [121]. Kerl and Staubesand [271] examined the effects of the coagulation of the rat femoral artery on the femoral nerve located within 3 mm. Currents were delivered by commercial monopolar and bipolar coagulation units and settings routinely used for coagulation. The effects were maximal on large neurons and consisted of lysis and vacuolization of myelin sheets and Schwann cells with associated axonal degeneration. The lesions extended to at least 2 cm along the nerve. The effects of bipolar coagulation were less severe and limited to the coagulation site. These were histopathological observations, without electrophysiological testing, although no gross motor impairment was noted [271].

92 2.1.5.7. Damage from repeated stimulations The possibility of inducing nerve damage by stimulation of the facial nerve during parotidectomy is often sited and therefore some surgeons ovoid it altogether. However, the development of several implants used in humans for functional electrical stimulation should put these concerns to rest. Sufficient is cite studies where the parameters for stimulation were examined: continuous stimulation of the cat sciatic nerve for 8 hours produces no detectable damage at stimulation frequencies below 20 Hz [342]. Above these frequencies, the damage is dependent in a linear fashion on both amplitude and frequency. The largest myelinated nerve axons are damaged with degeneration found in 1-3 % of the axons [4; 342]. The currents used are in the order of 0.5 to 5 mA, which covers the dynamic range of the current amplitude – contraction intensity curve of peripheral nerves [4] [199], and is above the stimulation currents of nerve monitoring devices [218].

2.1.5.8. Nerve toxic substances The list of potential neurotoxic substances is beyond our scope and could be found elsewhere [307; 464; 551].

2.1.5.9. Parotidectomy data Direct mechanical trauma. In an experimental study in rabbits, Patey and Moffat [411] were surprised to find that the nerve is quite resistant to direct trauma. In order to obtain about 50% of postoperative facial paralysis, the authors had to drop 75 to 100 times, on an exposed facial nerve, "the solid nozzle of a modified metal ear syringe… the weight that was allowed to fall was 16 g". Whether their conclusion that "the trunk and main branches of the facial nerve of the rabbit are resistant to degrees of direct mechanical trauma far beyond those to which the human facial nerve is submitted during conservative parotidectomy" is true, remains to be verified.

Cooling. In experiments on rabbits, Patey and Moffat [411] subjected the facial nerve to cooling (the degrees were unfortunately not specified) and found an incidence of 80% of facial paralysis.

Irrigation of the wound with substances that are toxic to nerves. To decrease the postparotidectomy recurrences, often occurring close to the incision, Patey used irrigation of the wound, before closure, with perchloride of mercury [411]. Their patients had an incidence of facial nerve deficits of 62% with irrigation, and 40% without irrigation [411].

93 The role of ischemia was alluded to by Patey [409], based on the work of Blunt [50] describing the vascularization of the facial nerve. Patey concluded that "ischemia is the main factor in the functional paralysis following parotidectomy, and minor stretch the smaller branches of the nerve an alternative or additional factor". Patey tried papaverine with little (?) changes, so he excluded vascular spasm.

It is difficult to draw a firm conclusion on the physiopathological basis of postparotidectomy facial paralysis, when branches are not sectioned. Causes that can be reasonably excluded include neurotoxic substances, electrical stimulation, cold, and ischemia. Neuropathy from toxic substances could be reasonably eliminated, since using similar techniques and materials in a series of operations by the same surgeon fortunately does not always result in paralysis! Damage from electrical stimulation and cold injury seems also improbable from the experimental animal data, although measures of facial nerve temperatures during parotidectomy have not been published. Similarly, facial nerve blood supply or oxygenation during parotidectomy has not been measured but seems unlikely in view of the rich vascular supply of peripheral nerves.

Facial nerve damage causes that should be recognized intraoperatively include electrocautery and crushing injuries. Unless careless grasping of the nerve branches occurs during parotidectomy, crushing injuries are probably an infrequent cause of facial nerve paralysis. The minimal safe distance for mono- and bipolar electrocoagulator use near nerves has not been established and it is unclear whether nerve damage can occur from electrocautery without visible facial movements.

The most probable causes of postparotidectomy facial paralysis without intraoperative problems are mechanical: compression and stretching. While the compression levels necessary to develop long-standing paralysis are rarely encountered during parotid surgery, nerve stretching of 10% can easily be imagined [137].

2.1.6. Incidence of postparotidectomy facial nerve paralysis

Facial nerve evaluation after parotidectomy has remained subjective until now. Only three reports have evaluated the postparotidectomy facial nerve function using an accepted facial nerve classification. These include the reports of Arndt et al. [14] using the Stennert scale, and the two reports using facial nerve monitoring (see § 2.1.7.): Olsen and Daube [399] and Wolf et al. [554]

94 using the House-Brackmann grading system. Of note is also the paper of Watanabe et al. [538] who used their one scale. All these publications are within the last 5 years.

The publications in the literature giving postparotidectomy facial nerve results are listed in Appendix I. A wide variation of the frequency of facial nerve paralysis can be found (Figure 20 and 21). Described risk factors for an increased facial nerve paralysis include:

1) More extensive surgery, i.e. more facial nerve deficits with total versus superficial parotidectomy [355; 401; 409; 411; 450; 536; 538], although some studies did not find such an association [367; 386]; 2) Previous parotid surgery, i.e. more paralysis in recurrent cases [355; 450; 536]; 3) Malignant tumors [61; 450; 538]; 4) Lesion size [450; 538], although other studies did nor find such an association [291; 367]; 5) Inflammatory conditions [61; 409; 507; 529].

The role of these factors in the frequency of temporary and permanent facial nerve paralysis, as found in the literature, is summarized in Table XIV.

Mra et al. [367] studied several factors in relation to postoperative paralysis in 65 patients with benign lesions and found only age as a statistically significant factor, although the population size might have been small. A similar conclusion was reached by Laccourreye and al. [291] on 229 patients with pleomorphic adenoma. Besides age, Terrell et al. [511] found operating time as significant factor in a multivariate analysis.

95

Variable Number of Temporary facial Permanent facial studies paralysis paralysis

Enucleation 21 9.5 ± 10 % 1.6 ± 2.1 %

Type of surgery Superficial parotidectomy 33 31 ± 24 % 4.4 ± 8.1 %

Total parotidectomy 29 42 ± 27 % 11.2 ± 14.7 %

NO 73 29 ± 23 % 5.6 ± 8.8 % Previous surgery YES 26 27 ± 18 % 16 ± 15 %

Benign 97 24 ± 21 % 7.8 ± 11.1 %

Histology Infections 3 53 ± 13 % 15 ± 24 %

Malignant 14 63 ± 9 % 27 ± 17 %

< 2 cm 2 35 ± 39 % 2.8 ± 4 %

Tumor size 2 – 4 cm 2 34 ± 41 % 6.0 ± 8.5 %

> 4 cm 2 67 ± 20 % 12.3 ± 21 %

Table XIV: Incidence of facial paralysis, both temporary and permanent in the literature This is a cross tabulation of the data from Appendix I regarding the frequency of facial nerve paralysis. The mean and standard deviation were calculated taking each study as an event. Extrapolation in terms of patient incidence might not be entirely correct because studies with small number of patients have an equal weight as larger studies. Because of this, the heterogeneity of the studies, and the lack of use of specific facial nerve evaluation scales statistical significance is not calculated. From Dulguerov et al. [132], without permission.

96 100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0% 1938 1963 1970 1972 1975 1976 1980 1984 1985 1988 1991 1992 1993 1994 1995 1997 1997

Figure 20: Incidence of temporary facial paralysis in the literature

97 50%

40%

30%

20%

10%

0%

0 4 5 38 8 8 8 88 91 92 93 94 95 97 97 9 9 1 1963 1970 1972 1975 1976 19 19 19 19 19 19 19 19 19 19 1

Figure 21: Incidence of permanent facial paralysis in the literature 98 2.1.7. Prevention of facial paralysis during parotidectomy

Very little has been written on preventing facial nerve injury during surgery. Again, the first author to speak specifically about prevention of postparotidectomy facial nerve paralysis was Patey [411]. Because of some experimental work on rabbits, Patey thought that the main cause of the "functional paralysis" was ischemia, so he recommended to avoid freeing the facial nerve trunk close to the stylomastoid foramen, in order to avoid injuring the stylomastoid artery which supplies the descending intratemporal segment of the facial nerve. In the same line of thought, he proposed limiting the complete freeing of the branches of the facial nerve to the strict minimum required by the tumor resection [411].

When this study began, there was no published report on facial nerve monitoring during parotid surgery. Olsen and Daube [399] were the first to publish about some form of facial nerve monitoring during parotid surgery. They used, apparently, a regular EMG machine with electrodes placed in the frontalis, orbicularis oculi and oris, and mentalis muscle. The monitoring was used in 7 patients operated on for a recurrent pleomorphic adenoma. While the immediate postparotidectomy facial status is not given, long term results were good with 5/7 patients having an HB grade of 1, 1/7 a grade 2, and the remaining patient had a grade 6 because of the sacrifice of the major portion of his facial nerve [399].

Wolf et al. [554] used two commercially available nerve monitors: the NIM-2 (Xomed, Jacksonville, Florida, USA) and the Neurosign 100 (The Magstim Company, Dyfed, United Kingdom). The use of intraoperative facial nerve monitoring in 35 patients was compared to a retrospective group of 24 patients. In the non-monitored group, the immediate post-operative paralysis was 75% (HB grades higher than 1), while in the monitored group it was 69%. Despite the lack of statistical analysis, the authors conclude that a "better functional outcome in the patient group with monitoring" was present. In addition, the immediate post-operative HB grades were somewhat better with the Neurosign apparatus, as compared to the Xomed one

99 2.2. Frey syndrome

2.2.1. Historical background

In 1923, Dr. Lucie Frey, a neurologist at the University of Warsaw, published her landmark paper on the "syndrome du nerf auriculotemporal" [185]. She described a 25-year-old patient who, five months after a gunshot wound to the parotid region, developed a facial sweating during meals.

.. le malade s'aperçut que, lorsqu'il mangeait, la moitié gauche de sa face devenait le siège d'une transpiration abondante accompagnée en même temps d'un vif sentiment de chaleur... Il était convaincu que cette transpiration anormale était mise sur le compte d'une trop grande voracité et il en avait honte. Dans la région qui correspond presque exactement au domaine du nerf auriculotemporal, on observe l'hyperesthésie de tous les modes de sensibilité. Lorsque le malade mange ou bien qu'il suce un bonbon, on note du côté gauche et au bout d'environ 1 à 2 minutes, une rougeur de la face, une élévation de la température locale et une abondante transpiration. La sueur, apparue d'abord sous forme de grosses gouttelettes, s'écoule après leur fusion en véritables rigoles. Les mouvements de mastication seuls, sans aliments, ainsi que l'excitation par le toucher de la muqueuse linguale ne provoquent aucun de ces troubles. Par contre, ils apparaissent chaque fois qu'on irrite la partie postérieure de la muqueuse linguale par des excitations gustatives et sans que les mouvements de mastication ou de succion interviennent. Lucie Frey, 1923 [185]

This paper presents a very precise description of the symptoms of the syndrome bearing her name. Frey syndrome is characterized by: 1) lateral facial sweating during meals; 2) a local facial skin flushing during meals. A local skin anesthesia was postulated by Frey [185], but is rarely observed. Once present, the gustatory sweating and flushing remains unchanged, i.e., there is no spontaneous resolution, even after numerous years. The area involved is on the lateral aspect of the face and upper neck, usually in the parotid region. Sometimes, the problem is localized behind the ear lobe, on a variable portion of the upper neck, in the hair bearing area in front or behind the pinna, or even inside the external auditory canal [434]. Also, there is sometimes a discrepancy between the area involved in the gustatory sweating and the one presenting a gustatory flushing [517].

Lucie Frey [185] not only correctly described the symptoms, but she also presented accurately the relevant autonomic innervation of the parotid gland and facial skin, pinpointing the role of the

100 auriculotemporal nerve in this syndrome. Synonyms of Frey syndrome include auriculotemporal syndrome [36; 102; 169; 183; 185; 193; 195; 375; 419], and gustatory sweating and/or flushing [51; 193; 226; 233; 284; 285; 286; 287; 320; 383; 533]. Whether the term "auriculotemporal syndrome" is correct, is questionable [228; 300], since the area involved, as previously discussed, often extends in the cutaneous distribution of other branches of the trigeminal nerve, as well as in the distribution of the greater auricular nerve from the cervical plexus. Nevertheless, this term is correct in terms of physiopathology.

In the literature, the first case of gustatory sweating is often credited [15; 75; 85; 193; 223; 228; 233; 257; 278; 279; 319; 370; 387; 473; 479; 484; 491; 497; 504][Black MJM, 1990 #967][Cliff S, 1998 #914] to M. Duphenix in 1757 [147] (Figure 22). The patient was injured by a deer during hunting and had a deep penetrating wound in the left cheek, through which a malar bone fracture could be appreciated.

Cette dernière playe, commençoit précifément fur l'angle de la mâchoire inférieure, pénétroit le corps du mufcle maffeter, et fe continuoit fous l'os de la pomette... Le gonflement étoit confidérable; toutes les parties voifines et le dedans de l'œil étoient pleines de fang extravafé. ...Du trente-neuvième au quarante-unième jour de la bleffure, il fortit plufieurs efquilles d'os, et peu de jours après, la playe fe réduifit à peu chofe en apparence: mais quand le bleffé commença à mouvoir fa mâchoire, il fortit par la petite ouverture qui reftoit, une grande quantité de falive, ce qui acheva de me perfuader que le canal de la glande parotide avoit été dechiré & rompu par le coup d'andouillet, ainfi que je l'avois annoncé dès les premiers jours. …Curieux de fçavoir ce qu'il perdoit comme falive dans un repas, je la fis recevoir dans un gobelet. La première fois que je fis cette épreuve, je trouvai qu'il s'étoit écoulé en quinze minutes, deux onces un gros de falive. Une feconde fois, en dix-huit minutes, il en fortit deux onces fix gros. Un autre jour en vingt- trois minutes, j'en reçus trois onces deux gros et demi. Enfin à la quatrième expérience, on ramaffa quatre onces un gros en vight- huit minutes. Duphenix, 1757 [147]

101 The wound remained open for more than four months during which clear liquid was observed whenever the patient was eating. After a surgical closure and the creation of drainage toward the oral cavity, the laterofacial liquid drainage appeared to have stopped. This is obviously a traumatic parotid fistula, and even the title of the paper suggests so (Figure 22) [147].

Figure 22: First page of Duphenix's 1757 manuscript

102 It is surprising that Nicolai, who wrote in 1985 a paper specifically reviewing this case [387], could interpret this description as a Frey syndrome. The presence of an open wound, the almost certainty that Stenson's duct was severed, since the masseter was visible through the wound, the short delay between the accident and the complication, the resolution of the symptoms, and the quantity of liquid pouring out during meals, all speak against a Frey syndrome [134]. The maximal output of a large Frey syndrome surface (7x7 cm ~50 cm2) is calculated (see § 2.2.5.) as 250μl/min. For 20 minutes, the maximal sweat output of such a surface is 5 ml. This is much less than the 56 to 102 ml that Duphenix could gather from his patient, notwithstanding possible evaporation and collection losses [134].

Recently [15; 319] credit was given to Dupuy [148], who was supposed to have described in 1816 "gustatory sweating over the cheek area in patients…". As the title of Dupuy's paper implies [148], it is an early experimental work on horses. While Dupuy's descriptions bear little relation with Frey syndrome, the work is quite interesting and probably represents one of the first studies examining the effects of sectioning the cervical sympathetics. A pretty clear description of the ocular signs are given, some 50 years before Horner described the symptom bearing his name [141; 240].

In 1853, Baillarger [19] described five cases and reviewed the case described by M. Duphenix. Two of the cases (II and VI) are typical salivary fistulae, one of long duration and the second one after drainage of an abscess. One of the cases (V) appears to be a parotid abscess, with symptoms lasting only a few days, which resolved after an incision and drainage; this complete resolution of symptoms is not seen in Frey syndrome. The two other cases (I and IV) are typical Frey syndrome, following surgical drainage of parotid abscesses.

La sueur commençait, comme je l'ai déjà dit, immédiatement après les repas, la peau de la face qui en siège devenait d'abord rosée, puis aussitôt elle se couvrait d'une multitude de petites gouttelettes limpides, brillantes, qui ne tardaient pas à se rassembler en grosses gouttes, et bientôt définitivement par ruisseler. La sueur coulait ainsi pendant une demi-heure à peu près, puis elle diminuait assez rapidement, en même temps que la peau pâlissait; enfin elle cessait complètement, la peau reprenait sa coloration naturelle, et tout rentrait dans l'état normal. M. Baillarger, 1856 [19]

There is little doubt that these patients had a typical Frey syndrome. However, probably because of Duphenix's interpretation, and because at the autopsy of one these patients Baillarger

103 felt that Stenson's ducts were blocked, he proposed that the fluid appearing during meals was an outpouring of saliva through the skin because of a blockage of Stenson's duct [19]. Nevertheless, we probably should regard Baillarger's publication as the first report of Frey syndrome [134; 524].

In 1849, Brown-Séquard discussed the possible production of facial sweating when eating spicy foods. He essentially described his own case of an exaggerated physiologic sweating, a interesting historical note that was largely echoed in the literature [28; 54; 185; 375; 385; 419; 431; 432; 549]. Nevertheless, the possibility of facial sweating is described, and this undoubtedly helped later authors such as Botkin [54] to correctly understand the nature of the fluid observed in the cases they have reported.

In 1859, Rouyer [449] gives brief descriptions of three cases of gustatory sweating. One case followed a bullet wound to the parotid area and is probably a typical Frey syndrome. In the other two cases, gustatory sweating is attributed to parotitis, although both had parotid abscesses. Most probably, as discussed below (§ 2.2.2.), these patients had their parotid abscess drained and developed a typical Frey syndrome. Unfortunately, following Baillarger [19], Rouyer also felt that the Stenson’s duct was blocked and that the saliva somehow found its way to the skin.

In 1875, Botkin, from St Petersburg, also described a case of Frey syndrome after drainage of a parotid abscess [54]. The case description is rather short and the interpretation ambiguous: although the symptoms are maximal on the cheek and during meals, apparently the entire half of the body can be involved and, eating, as well as walking, excitement etc. could induce the symptoms.

In 1888, Paul Raymond wrote two seminal papers on the “Ephidroses de la Face” where the role of the autonomic sympathetic system in the development of cutaneous flushing and sweating appears established [431; 432]. One of the cases described has a Horner’s syndrome with gustatory facial sweating and flushing. This is one of the first reports of gustatory facial sweating and flushing following a lesion of the cervical sympathetic chain.

The first case of Frey syndrome in the English literature was described by Weber [543] in 1897. It is also the first case of bilateral Frey syndrome. The patient had bilateral scars in the parotid area, after incision and drainage of "previous suppurations." Like Baillarger, the clinical description is exact with flushing, sweating, a clear relationship with gustatory stimuli, and an absence of effect of chewing.

104 In 1922, a year before Frey's description, New and Bozer [385] reported on three cases of "Hyperhydrosis of the cheek associated with injury of the parotid region." Two of the patients had drainage of a parotid abscess and the third one had a traumatic injury to the pre-auricular area.

Finally, Peter Bassoe reported the first case of Frey syndrome following parotidectomy in 1932 [28]. Once the syndrome was better known, several reports followed and, by 1948, at least 35 cases are published in the literature [183].

2.2.2. Etiology

While Frey syndrome is most often seen after parotidectomy, it has been described with other surgical procedures in the parotid region such as mandibular surgery [280; 282; 523][Guerrissi J, 1998 #861], drainage of parotid or other local abscesses [19; 195; 300; 385; 415; 419][Swanson KS, 1991 #971] and possibly neck dissection [370; 491]; with regional trauma such as mandibular fractures [198; 385; 433; 569][Martis C, 1969 #923][Mellor TK, 1996 #925][Laws IM, 1967 #968][Storrs TJ, 1974 #969][Dhaif G, 1995 #970][Gerbino G, 1997 #917] or penetrating wounds [15; 147; 433]; and with surgery or lesions of the cervical sympathetic chain [110; 226; 320; 383; 549].

Often in the pre-antibiotic literature, Frey syndrome is described as following "parotitis." A closer look at most of these publications shows that all described cases had had drainage of a parotid abscess [19; 195; 364; 385; 415; 419; 543], although in the publications various etiologies have been described. No recent case of Frey syndrome resulting from parotid infection has been reported.

More bizarre etiologies for facial gustatory sweating, reported in single cases, include herpes simplex of the mandibular division of the trigeminal nerve [130] and a case of a large cerebellopontine angle tumor [467]. In both of these cases, skin anesthesia in the involved area was present [130; 467]. These cases can also be explained by the aberrant regeneration theory (see § 2.2.4, below), with the locus of cross fiber regeneration being at the level of the maxillary division of the trigeminal nerve or even in the deep portion of the auriculotemporal nerve instead of the parotid space portion of the auriculotemporal nerve. A similar explanation, i.e., peripheral parasympathetic sprouting to degenerated sympathetic fibers [368], has been offered for the observed facial, and often neck and upper chest, gustatory sweating seen in patients with severe diabetic neuropathy [184; 227; 254; 501; 540].

105 Haxton related that Claude Bernard had a spontaneous, idiopathic gustatory sweating when eating chocolate [226] and Frey [185] and others [28; 375; 385; 431; 432; 549] have echoed Brown- Séquard’s own case report [66] of similar problems when eating spiced foods. Whether these truly represent a Frey syndrome, or simply a normal variant, remains to be seen, since cutaneous facial vasodilatation and sweating can be regularly elicited with very spicy foods such as chilies [308].

The few cases of pediatric Frey syndrome that have been described [22; 36; 116; 169; 257; 280; 484][Cliff S, 1998 #914] merit careful review because they possibly represent the only etiology, not easily explained by the aberrant regeneration theory. The majority of the cases described are normal children with uni- [22; 36; 116; 169; 280; 484][Cliff S, 1998 #914] or bilateral gustatory facial flushing [116; 257], without any sweating. First, some cases in children are typical Frey syndromes following surgery to the parotid gland [285], or drainage of parotid abscesses [385; 419]. Hopefully, similar details in the patient's history were not missed in the other cases by authors with little exposure to typical Frey patients. Second, facial sweating is the hallmark of Frey syndrome, and the majority of reported cases are probably a normal variation and, in our opinion, should not be regarded as a classical Frey syndromes, and even less as "auriculotemporal syndromes [22; 36; 169; 257; 484]. Maybe the term "gustatory flushing syndrome," as proposed by Kozma and Gabriel [280], might be more appropriate. Third, in the few cases with a follow-up of several years, a gradual resolution of the symptoms occur [116; 280; 357], in contradistinction of typical Frey patients. Fourth, the possible role of traumatic forceps delivery, pointed out by Balfour and Bloom [22], remains to be demonstrated. However, the unilaterality of the symptoms in several cases and the large number cases with forceps delivery [22; 36; 116; 257; 484][Cliff S, 1998 #914] could point to a local trauma as a possible cause of the syndrome.

In addition, a "hereditary" form has been described in one report [328]. "Gustatory otorrhea" in a single patient without any previous pathology has also been recently described [434].

106

Figure 23: Autonomic innervation of the parotid gland and facial skin – overall view From Wilson-Pauwels AC, Akesson EJ, Stewart PA [550] without permission.

107 2.2.3. Anatomy and physiology

During eating, a reflex arc is stimulated leading to increased salivary secretion of minor and major salivary glands, including the parotid gland. The origin of the afferent limb is at the level of the gustatory papillae of the oral cavity and oropharynx. The afferent neurons terminate in the nucleus solitarius and project to the inferior salivary nucleus. Pre-ganglionic parasympathetic efferent fibers leave the inferior salivary nucleus and the brain stem with the glossopharyngeal nerve (Figure 23). After passing through the jugular foramen and the superior and inferior glossopharyngeal ganglions, the parasympathetic fibers follow the tympanic nerve (Jacobson's nerve). Jacobson's nerve enters the temporal bone and travels in the inferior tympanic canaliculus towards the middle ear cavity. A plexus is formed on the promontory, but the parasympathetic fibers not supplying the middle ear unite to form the lesser superficial petrosal nerve (Figure 24). This nerve exits the temporal bone and travels on the floor of the middle cranial fossa. The fibers then exit the skull through the foramen ovale and reach the otic ganglion where they synapse with post-ganglionic parasympathetic fibers. These post-ganglionic parasympathetic fibers then join the auriculotemporal nerve and reach the parotid gland through its medial aspect in the parapharyngeal space [60]. Like all post-ganglionic parasympathetic neurons, the neurotransmitter of these fibers is acetylcholine.

The sympathetic fibers for the parotid gland originate from the cervical sympathetic chain. The postganglionic sympathetic fibers arise in the superior cervical ganglion and follow the periarterial plexuses of the internal and external carotid arteries.

The secretion of the parotid gland is controlled predominantly by the parasympathetic system. In cats, stimulation of the parasympathetic fibers produces an important increase of salivary output, while sympathetic stimulation results only in a discrete increase [441]. In human volunteers undergoing stapedectomy for otosclerosis, Ross [447] measured the salivary production during electric stimulation of the tympanic plexus. The quantity of saliva increased progressively with the intensity of the applied electric current. Stimulation of the chorda tympani did not increase parotid salivary output.

The sympathetic fibers for the cutaneous vessels and sweat glands follow the cervical sympathetic chain and a synaptic relay is found in the sympathetic cervical ganglions. The fibers then follow the periarterial plexuses of the internal and external carotid arteries and their branches [320; 337]. By analyzing detailed clinical observations, List and Peet [320] conclude that these fibers eventually join and follow the branches of the trigeminal nerve supplying the cutaneous sensation

108 of the skin. For the ophthalmic division the sympathetic fibers join the nerve before it exits the skull, while for the maxillary and mandibular division the anastomosis is extracranial, close to the foramen ovale and rotundum [320]. While, the overall organization is described in anatomical textbooks, the exact pathway is still debatable [541].

The sympathetic supply to the submandibular gland follows branches of the external carotid artery [337]. At some point of the lower face or upper neck region, a switch from sympathetic fibers traveling with the internal carotid artery to sympathetic fibers traveling with the external carotid artery takes place [193; 337; 465; 541]. It is also possible that there is some overlap with double sympathetic innervation along the trigeminal nerve and along the external carotid artery [465][Drummond PD, 1987 #966].

In contradistinction to other postganglionic sympathetic fibers, which are adrenergic, sweat glands have acetylcholine as the main neurotransmitter, as shown initially in the cat's foot [112] and in human facial skin [180; 308]. Despite the more recent demonstration of catecholamine [525], vasoactive intestinal peptide [328], atrial natriuretic peptide, calcitonin gene-related peptide, and galanin [509] in periglandular nerves, the functional role of these peptides is uncertain [461].

Figure 24: Autonomic innervation of the parotid gland and facial skin - detail From: Wilson-Pauwels AC, Akesson EJ, Stewart PA, [550] without permission.

109 2.2.4. Pathogenesis of Frey syndrome

When an irreconcilable factor is present, any hypothesis must be regarded with suspicion, and the explanation given here may have to be modified if further facts come to light. Haxton, 1948 [226]

A number of different theories have been proposed to explain Frey syndrome. Frey, herself, proposed that the symptoms resulted from an irritation of the sympathetic fibers traveling, within the auriculotemporal nerve, through the parotid bed. The presence of scar tissue in the parotid area somehow irritated these sympathetic fibers during periods of parotid stimulation [185]. Essentially, what is postulated is an exaggeration of a normally occurring physiologic reflex [375].

Despite an excellent description of the anatomy and astute observations that greatly contributed to the understanding of the syndrome, List and Peet [320] also proposed an irritation explanation: irritation of parasympathetic fibers of the auriculotemporal nerve. The remaining parasympathetic fibers and the scarred parotid gland "maintain a local condition of abnormal irritability of the cholinergic fibers" [320].

Haxton [226] followed Langenskiold's [300] suggestion that there is a local hypersensitivity of cutaneous target organs to acetylcholine. Haxton's conclusion is based on the experimental abolishment of the gustatory sweating in patients who had undergone a cervical sympathetectomy by low cervical infiltration of a local anesthetic (procaine). These findings have yet to find a definitive explanation, but one possible reason is that the drugs were systemically absorbed. Haxton could not find a reasonable source for the acetylcholine causing the gustatory sweating [226]. Langenskiold suggested that acetylcholine originated from the parotid gland [300]. This hypersensitivity theory was also supported by Freedberg et al. [183] despite their excellent anatomical and physiological investigation of the syndrome.

Andre Thomas [517] and later Ford and Woodhall [179] postulated that the severed parasympathetic fibers to the parotid gland regenerate along the wrong neurilemmal sheaths of the sectioned cutaneous sympathetic fibers, the so-called aberrant regeneration theory. Ample evidence for the existence of collateral sprouting and heterogeneous regeneration in the somatic and autonomic nervous system is available [215; 368]. It might also be remembered that the source for nerve growth factor in Levi-Montalchini's experiences came from mice salivary glands [99; 316].

110 Several experimental and clinical observations support the aberrant regeneration theory:

1) Auriculotemporal nerve bloc by local anesthetics induces a local anesthesia and an abolishment of gustatory sweating of similar cutaneous distribution [183; 193];

2) Anesthetic infiltration in the region of the otic ganglion abolishes the symptoms [193];

3) Stimulation of the tympanic plexus results in sweating in the involved skin area [447];

4) Symptoms are unaffected by anesthesia of the stellate ganglion, despite the appearance of a temporary Horner syndrome [129; 183; 193];

5) In the involved area, there is an impairment of the normal sympathetic function, such as reduced thermoregulatory sweating [183; 193; 226; 320][Drummond PD, 1987 #966], absence of emotional flushing [183; 226; 320][Drummond PD, 1987 #966], or Horner's syndrome [226; 320];

6) A latent period (minimum 1-2 months [286]) is present between the initial injury and the appearance of gustatory flushing and sweating [193; 226; 319; 320; 549]. If patients are followed sequentially with an appropriate sweat test, the positivity of the test increases and the involved area progressively increases in size [286; 319];

7) The symptoms can be abolished by the local injection of anticholinergic substances such as atropine [183; 193];

8) A localized hypersensitivity to cholinergic drugs is described with mecholine [549], acetylcholine [183; 193], and pilocarpine [193; 549];

9) The syndrome appears after various lesions or resections of the homolateral cervical sympathetic chain [226; 320; 383; 549][Drummond PD, 1987 #966].

The postulated etiology is an aberrant regeneration of the sectioned parasympathetic fibers normally innervating the parotid gland (Figure 25). The traumatized fibers loose their parotid targets and regenerate to innervate the vessels and sweat glands of the overlying skin. In order for this aberrant regeneration to occur, the sympathetic fibers to these vessels and sweat glands have to be damaged, a frequent event in either parotid surgery or penetrating parotid trauma. The regular function of the parotid parasympathetic fibers is to increase salivary secretion during eating. The activation following aberrant regeneration produces an activation of the new targets during meals, resulting in a local vasodilatation ("gustatory flushing") and localized sweating ("gustatory sweating").

111

Sweat glands

Ski n

Pa ro t i d

Sy m p Psy m p

??? ATN

Sweat glands

Ski n

Pa ro t i d

Sy m p Psy m p

??? ATN

Figure 25: Schematic representation of the aberrant regeneration theory TOP: normal situation. BOTTOM: situation after superficial parotidectomy. Symp = sympathetic innervation of sweat glands; Psymp = parasympathetic innervation of the parotid and facial skin. ATN = Auriculotemporal nerve; ??? = Sympathetic nerve, the exact pathway of which is still unclear (auriculotemporal nerve, great auricular nerve, perivascular plexuses). From Dulguerov et al. [139] without permission.

112 A certain amount of "physiologic" gustatory sweating and flushing can be induced by certain spicy foods such as chilies in normal subjects [308]. A variation of the threshold of individual cutaneous regions appears to be present, since only certain head and neck areas are found to respond in a given individual [308]. The distribution was found to be similar to head and neck sweating and flushing induced by elevation of the ambient temperature. Also, the threshold for gustatory stimuli is decreased by increasing the ambient temperature. In contradistinction to Frey syndrome, this "physiologic" gustatory sweating and flushing is bilateral and usually symmetric [308], similar to physiologic thermal sweating [427]. The "hereditary gustatory sweating" described by Mailander [328] is probably an exaggerated form of physiologic sweating.

2.2.5. Investigation of Frey syndrome

Non-invasive tests of autonomic function are easy to do but hard to interpret. P.A. Low, 1993 [323]

Since the most troublesome symptom of Frey syndrome is sweating, testing for Frey syndrome has in general been limited to sweating. Only Laage-Hellmann investigated flushing, and this was done by simple direct observation and without any quantification [284; 286].

Therefore, testing for Frey syndrome is assessing the function of eccrine sweat glands. The major function of sweat glands in humans is thermoregulation [458; 461]. About 2-4 million sweat glands are distributed over the entire human body with a density varying from 60 glands/cm2 on the back to 600 glands/cm2 on the palms and soles [460]. The maximal sweat rate is about 2-20 nl/min/gland [461]. The face has about 250 glands/cm2 and thus the maximal facial sweat rate should be about 0.5-5 μl/min/cm2 or 5-50 ml/min/m2. Assuming that the maximal sweat secretion rate is observed in Frey syndrome and that a large surface is involved, for example 7x7 cm or 50 cm2, the maximum amount of sweat liquid that can collect is 250 μl/min. Sweat is a hypotonic solution that contains mainly sodium chloride, potassium and bicarbonate, as well as certain inorganic compounds such as lactate, urea, and ammonia [458; 461].

Tests of autonomic thermoregulatory function are complex [323] and the measure of sweat output is just one of the aspects assessed. Tests of sweat output function measure the production of sweat by a group of sweat glands and these tests could be classified as: 1) topographic, where a chemical reaction provides a view of the anatomical location of sweat secretion; 2) electric, where a

113 change of skin impedance by the humidity of sweat is measured; and 3) thermodynamic, where the skin humidity is evaporated and a "sudometer" measures the thermal mass of the air stream [323]. Advantages of the electric and thermodynamic tests include the possibility of repeated and dynamic measures, while topographic tests give a better representation of surface and anatomical distribution.

The ideal test method for Frey syndrome should provide topographic information and possibly quantification of the amount of sweating. While dynamic measures might be interesting in investigation studies, their use in clinical practice is of limited value. Further characteristics of the test should include: 1) simplicity (one-step) of the method; 2) sensitivity, so that different sweating rates could be appreciated; 3) absence of toxicity and allergenicity of the agents used; 4) easy removal from the skin of the applied agents; 5) low cost [460].

The most frequently used method of sweat secretion assessment for Frey syndrome [46; 51; 164; 183; 228; 233; 249; 278; 279; 284; 285; 286; 287; 302; 308; 319; 320; 328; 338; 370; 373; 375; 469; 473; 479; 524; 562][Black MJM, 1990 #967[Mellor TK, 1996 #925][Cliff S, 1998 #914][Guerrissi J, 1998 #861][Sood, 1999 #933] was originally described by Victor Minor, a Russian neurologist [360]. A solution containing 1.5 g iodine, 10 g castor oil, and 88.5 g of absolute alcohol is painted on the skin. After drying, the areas are powdered with starch. The water in the sweat produces blue coloring by a reduction reaction of the iodine-starch mixture. With limited sweat production, the aperture of individual sweat glands are marked as small blue dots, while with larger amounts of sweat secretion the blue dots become larger and eventually become confluent. The Minor test is a topographic method allowing accurate mapping of the involved surface. Pictures can be taken and have been extensively published. Disadvantages include the necessary application of several layers of regents, the difficulty of removing the iodine paint, the possible allergy to iodine, the difficulty of using the method with heavy perspiration, and the lack of dynamic testing [462]. The Minor test is quite cumbersome and, not infrequently, patients tend to refuse it [39].

Modifications of Minor's test include the use of other dyes such as bromophenol blue powder [234], pyrogallol [564], and quinazarin [217]. Because of the length of the preparation and the necessity to restart the process for each measure, these methods cannot be used for dynamic measures. Furthermore, they are not well suited to provide a quantitative estimation of sweat output.

114 Randall [427], noting that some papers have starch coatings, modified the method by using paper instead of starch, but continued to paint the skin with the iodine-based mixture. The iodinated starch paper imprint method [126] further simplifies the process by the use of regular paper that has absorbed the iodine. The "iodine-coated" paper, like the Minor iodine-starch mixture, turns blue when wetted, and the paper is like a print, giving a mirror image of the sweating skin [144]. The iodinated starch paper imprint method is sensitive and well calibrated to provide quantitative results [126]. Probably the only major disadvantage of the iodinated-starch paper imprint method in testing Frey syndrome is the lack of dynamic results.

Another topographic test is the silastic imprint method, where the liquid produced by each sweat gland generates a void in the silastic mold. The void dots can than be counted and their volume estimated, if the image is digitized. While this technique is sensitive, no topographic imaging is provided, the test is not dynamic, and the equipment involved is often expensive and cumbersome to use [269; 323].

Laccourreye et al. proposed recently the use of a L-lactate skin electrode for the measurement of sweat gland output [289]. This technique appears interesting since lactate concentration in sweat is 10-20 times that of plasma [458; 461], no signal is detected in normal skin, and dynamic results can be obtained. Serious disadvantages for its application in Frey syndrome are that no topographic data are generated and that the equipment involved is expensive and complicated.

115 2.2.6. Incidence of Frey syndrome

The incidence of Frey syndrome is variable. The publications of Laage-Hellmann [284] in the late 1950’s constitute the first serious attempt to study this problem. A group of 123 patients who underwent a parotidectomy were evaluated retrospectively with a questionnaire and clinical testing. On questioning, 62% of the patients complained of problems during eating: 22% had both sweating and flushing, 26% had sweating only, and 14% had only flushing. When a gustatory stimulus was used, a flushing was observed in 92% of patients, and 98% of patients exhibited a positive Minor test (Figure 26).

Frey syndrome after parotidectomy (Laage-Hellmann, 1958)

120%

98% 100% 92%

80%

62%

% 60%

40%

20%

0% Positive history Observed flushing Positive Minor test

Figure 26: Frequency of Frey syndrome following parotidectomy Positive history = patients who on questioning complained of alimentary sweating or flushing. The gustatory stimulus for the investigation of flushing and for the Minor test [360] was a slice of lemon sucked for 2 minutes. Data from Laage-Hellman [284].

Laage-Hellmann [284; 285; 286; 287] concluded that Frey syndrome was an unavoidable sequel of parotidectomy that is not overtly symptomatic in all patients. This might be a severe judgment, but an unavoidable conclusion is that the patient's history is an unreliable assessment of the incidence of Frey syndrome. Kornblut et al. found a positive history in 33% of his patients and a positive Minor test in 96% [278].

116 In Laage-Hellman's study [284], the severity of the problem was judged moderate by 58% of patients, important by 15% and embarrassing by 27%. When these percentages are reported for the entire group, the numbers are: moderate 28%, important 7%, and embarrassing 13% (Figure 27). The correlation between the severity of sweating and intensity of the Minor test was not good. Similar results are found by Kornblut et al. [278; 279].

Severity of sweating

80%

60%

58%

% 40%

28%

20% 27% 13% 15% 7%

0% Moderate Important Embarassing

Patients w ith a sw eating history All patients

Figure 27: Severity of gustatory sweating symptoms Patients with Frey syndrome after parotidectomy were asked to qualify the severity of their symptoms. Data from Laage-Hellman [284].

Laccourreye et al. [289], using a L-lactate skin electrode for the measurement of sweat gland output, concluded that the clinical severity of Frey syndrome was related to the surface involved. A somewhat better correlation is found between the extent of the parotid surgery and the intensity of the Minor test, with less abnormal scores associated with less extensive surgery (Figure 28). This is probably only valid for enucleation or the limited superficial parotidectomy.

117 Minor test scores according to type of operation

100%

90%

80%

70%

60% 3 2 50% 1 40% 0

Number of patients 30%

20%

10%

0% Partial superficial Conservative superficial Subtotal parotidectomy parotidectomy parotidectomy

Figure 28: Minor-test scores in different parotid surgery groups 0 denotes a negative Minor test, 1 is a slightly positive Minor test, 2 is a moderately positive Minor test, 3 is a very positive Minor test. Data from Laage-Hellman [284].

The delay between parotid surgery and the appearance of Frey syndrome was examined by sequential Minor tests by Laage-Hellman [286]. The minimal delay was 5 weeks and the median delay for a positive Minor test was around 8 weeks. The test was positive before patients started to complain about gustatory sweating. In some patients, the test became positive at 9 months and in only 1 patient did Frey syndrome appear at 1 year (patient who had postoperative radiotherapy). Other authors [228] have reported the clinical manifestation of Frey syndrome up to 8.5 years after parotidectomy.

The incidence of Frey syndrome in other publications was not as thoroughly investigated. An overall view is presented in Figure 29. While the incidence of clinical Frey syndrome is variable, the incidence of objective Frey syndrome has always been high, around 80-90% (Table XV).

118 Incidence of Frey syndrome

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0% 1947 1956 1960 1968 1975 1976 1980 1982 1987 1988 1991 1991 1992 1992 1993 1994 1996 1996 1997 Frey - clinical Frey - tested

Figure 29: Incidence of Frey syndrome in the literature

119

Author Year Number of Incidence of clinical Frey Incidence of objective patients syndrome Frey syndrome

Laage-Hellman [284] 1958 123 62 % 98 %

Kornblut et al. [278] 1974 35 43 % 97 %

Gordon & Fiddien [201] 1976 50 34 % 100 %

Farrell & Kalnins [164] 1991 21 14 % 42.9 %

Yu & Hamilton [562] 1992 35 5.7 % 14.3 %

Allison & Rappaport [7] 1993 35 83 % 87 %

Linder et al. [319] 1997 193 23 % 93 %

TOTAL / AVERAGE 492 38 % 87 %

Table XV: Incidence of Frey syndrome in the literature using objective testing From Dulguerov et al. [138] without permission.

2.2.7. Treatment of Frey syndrome

Several treatments have been advanced for Frey syndrome. In Frey's publication [185], she mentioned that a surgeon had infiltrated alcohol around the scar with a temporary relief of the patient's symptoms.

The treatment modalities used can be classified in five main groups: a) section of some portion of the efferent neural arc; b) external radiotherapy; c) local or systemic application of anticholinergic drugs; d) interposition of a subcutaneous barrier, as used for Frey syndrome prevention; and e) injection of botulinum toxin in the involved skin. Considering that the symptoms are not always very troublesome, it is important that the treatment itself does not have more side effects.

A) Section of the auriculotemporal nerve was originally described by Leriche for the treatment of traumatic parotid fistula [313]. Coldwater attempted the procedure for Frey syndrome in 2 patients under local anesthesia, with only partial and temporary decrease of the symptoms [102]. Similarly, Patey mentioned that "we have recently been trying the effect of prophylactic auriculotemporal nerve avulsion at the time of parotidectomy, but, though we are not yet in a

120 position to give full results, one case of gustatory sweating has already occurred" [193]. More recent series confirm the inefficacy of auriculotemporal avulsion [119]. Also, the access to this deeply located region is complex and carries the risk of injuring several important structures, including the facial nerve.

At the other end of the efferent pathway, a division of the intracranial portion of the glossopharyngeal nerve was done by Gardner and McCubbin [188], with a good result lasting 5 years in one patient and a short period of relief in a second one. Considering the extensive surgery, this therapy was not pursued any more.

Hemenway [233] suggested in 1960 to interrupt the efferent neuronal pathway at the level of the middle ear, by sectioning the tympanic nerve of Jacobson. The procedure was initially proposed by Lempert [310] for the treatment of tinnitus. The first such procedure for Frey syndrome was carried out by Golding-Wood, who named it "tympanic neurectomy" [197] in three patients with apparently excellent results. Ross's [447] experience is less enthusiastic: all patients were initially improved, but, with a follow-up greater than one year, three patients out of five had a recurrence. An anatomic temporal bone dissection revealed that Jacobson's nerve divides rapidly after its entry in the hypotympanum and a complete section is not easily performed [223; 447]. Hays [228] reviewed the literature up to 1978 and found 73 patients who had undergone tympanic neurectomy: 56% had excellent results, in 26% the results were satisfactory, and 18% unsatisfactory (follow-up not always specified, objective test not used).

In summary, the only valid surgical treatment of Frey syndrome appears to be tympanic neurectomy. The procedure appears to be based on solid physiopathologic grounds, does not seem too risky, can be done under local anesthesia, and probably on an outpatient basis. Unfortunately, it is really unclear whether the effectiveness of this procedure is long lasting. Few reports on tympanic neurectomy have been published in the last 20 years. Stevens and Doyle [497] describe a case of bilateral Frey syndrome, which required several bilateral tympanic neurectomies, that not only failed ultimately, but also resulted in bilateral gustatory rhinorrhea. Recently, Linder et al. [319], discussing the treatment of Frey syndrome, mention that 3 tympanic neurectomies were performed in Zurich between 1970 and 1990: a recurrence of symptoms within 12 months was present in 2 patients (no objective facial sweating testing was performed).

B) Radiotherapy was attempted by Needles [375] in one patient, but the results were unsatisfactory since his patient interrupted the radiation. Some of Laage-Hellman's patients were

121 irradiated postoperatively and 2 out of 5 patients did not develop Frey syndrome [284]; also, the gustatory induced sweating remained outside of the radiotherapy fields [286]. Similar hints can be drawn from other publications [13; 118; 164; 319].

C) The attempts to treat Frey syndrome with systemic anticholinergic medications [375] were put to rest by the paper of Shelley and Horvath who showed that none of substances available at that time could be used in accepted doses to reduce the sweating produced by 0.1 cc intradermal injection of pilocarpine [475]. These authors also showed that the local iontophoresis or application of anticholinergic drugs such as scopolamine could be used to reduce sweating [475]. Laage-Hellman was the first to apply scopolamine for the treatment of Frey syndrome [285]. The application of 3% scopolamine cream reduced the gustatory sweating in the 11 male patients and abolished the symptoms in the 6 female patients, a difference tentatively explained by a thinner female skin. Six of these 17 patients noticed side effects such as dryness of the mouth and ipsilateral ocular problems. Hays [228], in a double-blind, self-controlled study in 16 patients, evaluated locally applied glycopyrrolate (1%): in all patients the symptoms were reduced or abolished for a duration of 4-10 days; no side effects were noticed. May and McGuirt [338] confirmed these findings in 5 patients. Huttenbrink used aluminum chloride in 28 patients with good results [246]. Laccourreye et al. [290] used a similarly designed study in 15 patients for the evaluation of diphemanil methylsulfate: complete relief was observed by 40% and partial by 33% of patients. Two patients complained of dryness of the mouth. Others have reported on various topical anticholinergic preparations in non-controlled studies of small size [Bednarek J, 1976 #911][Black MJM, 1990 #967].

In summary, it appears that topically applied anticholinergic drugs can be used to control the gustatory sweating and flushing with relatively limited systemic side effects. The cream or lotion needs to be applied only 1-2 times per week. Several precautions are necessary before such a treatment is undertaken (Table XVI). There is, however, only one publication available on patients treated with these drugs for longer periods [229]. A total of 22 patients used these drugs for periods from 6 months to 4 years. In 5 patients, the symptoms appeared to decrease with time and they discontinued the use of the medication. In 4 patients, higher doses of glycopyrrolate were necessary to achieve similar results. The side effects remained mild with one episode of dry mouth and one episode of ocular problems [229]. Despite the effectiveness of these preparations, patients seem, in the long run, to abandon the cumbersome daily application [319].

122

1. Symptomatic patient requesting treatment.

2. Topographic sweat test to demonstrate and delineate the location to be treated.

3. Intact healthy skin in the area to be treated. Instructions never to apply substance on cut or infected skin.

4. Ophthalmologic consultation to rule out glaucoma and narrow angle anterior chamber.

5. Medical consultation to rule out prostatic hypertrophy, pyloric obstruction, irritable bowel syndrome, asthma etc.

6. Storage of medication in child-resistant safety containers. Keep medication out of reach of children.

7. Rub cream preparation into skin with one finger for 20 seconds. Apply preparations in a solution with a cotton tip applicator.

8. Avoid instillation in the eyes, mouth, and nose.

9. Reapply after one day of recurrence of sweating and/or flushing.

10. Discontinue medication if dry mouth or other systemic side effects occur (physician consultation).

Table XVI: Guidelines for topical treatment of Frey syndrome with topical anticholinergics Modified after Hays [228]

D) Sessions et al. [473] pioneered the use of a barrier between the facial skin and the parotid bed in the treatment of post-parotidectomy Frey syndrome. They used the surgical interposition of a fascia lata graft in 4 patients with excellent results. Two of these patients had a follow-up of several years and were retested with the Minor test. Wallis and Gibson [533] published a similar case with a rather short follow-up of 8 months. More recently, Webster [Webster K, 1997 #936] reported on two patients treated with an interposition of lyophilized porcine dermal graft. This technique appears to be working in the small group of patients reported, but probably a major deterrent of its use is the risk of facial nerve paralysis, since the nerve lies just under the skin after superficial or total parotidectomy.

E) A new and interesting treatment was introduced by Drobik et al. [128]: the injection of botulinum A toxin in the skin involved by Frey syndrome. One patient had intradermal injections in the involved skin area with resolution of the symptoms lasting at least 12 months. A Minor test confirmed the absence of Frey syndrome after the injection. Similar results have also been reported by Schultze-Binhage et al. [469] in 3 patients, Quinodoz et al. in 6 patients [425] and Dulguerov et al. [143] in 15 patients.

123

2.2.8. Prevention of Frey syndrome

One of the first attempts to prevent the development of Frey syndrome was done by Kidd, who tried to resect the auriculotemporal nerve during total parotidectomy, as far proximally as possible [273]. The incidence of Frey syndrome following avulsion of the nerve was 20% and without avulsion 26% (Table XVII).

All other techniques described for the prevention of Frey syndrome are surgical and involve the placement of a barrier preventing the regenerating parasympathetic fibers from the parotidectomy bed to the skin vessels and sweat glands. The different techniques include pedicled muscular flaps, pedicled fascia, non-vascularized tissue such as fascia and dermal, microvascular grafts, and synthetic materials.

The first to suggest the use of a surgical interposed barrier was Kornblut et al. [278] in 1974. They used a superiorly based sternocleidomastoid (SCM) muscle rotation flap [526] that was rotated in the wound under the skin. This was initially used by Jost et al. in 1968 [264] in order to palliate the postoperative retro-mandibular depression that is sometimes quite unaesthetic (see section 2.3.) Kornblut's results [278] were rather disappointing since the incidence of Frey syndrome was higher in the SCM group, both on questioning and on testing with the Minor starch test. Casler and Conley [85] reported recently on the use of this technique for Frey syndrome prevention. They found that 47% of controls and 12.5% of the SCM group patients had a positive history of gustatory sweating, a statistically significant difference. Since Casler and Conley did not use any objective test of Frey syndrome, their results certainly underestimate of the frequency of the problem. More recently, Sood et al. [Sood, 1999 #933] reported good results in 11 correctly evaluated patients, with this technique.

Another rotational flap used is the temporoparietal fascia flap, proposed recently by Sultan et al. [504]. They extended the preauricular limb of the parotidectomy incision in the temporal hair and isolated the fascia covering the temporalis muscle. This fascia was then rotated over the parotidectomy bed. They used this method in 7 patients without clinical Frey syndrome; the follow-up is not stated.

The use of subcutaneously implanted non-vascularized dermal grafts was first proposed by Cassady in 1977 [86]. He placed such grafts after parotidectomy in dogs and used the controlateral side as control. Since even the control side did not develop sweating, the results are difficult to interpret. Nosan et al. [391] used abdominal dermal-fat graft in 9 patients: no gustatory sweating

124 developed but the method of assessment and the duration of follow-up were not stated. Harada et al. reported their results in 7 patients, followed for more than 2 years, and found only one case of clinical gustatory sweating at the edges of their graft; however no objective testing was used [222].

In 1980, Singleton and Cassisi [479] compared two surgical techniques of raising the skin flap: a thin flap technique, where the dissection is just under the hair follicles, and a deeper dissection, which they called thick flaps. Thin flap resulted in 12.5% incidence of Frey syndrome and thick flaps in 3% (p = 0.02). Evaluation was by history only and the minimal follow-up was only 8 weeks.

The superficial musculo-aponeurotic system (SMAS) of the face was originally described by Mitz and Peyronie in 1976 [361] and applied soon in rhytidectomy surgery [403]. Rappaport and Allison [428] were the first to apply the concept of SMAS coverage at the end of a parotidectomy. Their original goal was to decrease the retro-mandibular depression and, because of a short follow- up, they did not analyze Frey syndrome. In a recent publication [7], using both a questionnaire and the Minor test, they report a 1% incidence of positive history and a positive Minor test in 2% of patients undergoing SMAS coverage. Another group of patients without SMAS flap had a 83% incidence of positive history and a positive Minor test in 87% of patients. Bonanno and Casson [51] reported in 1992 their results with the SMAS technique in 55 patients: the incidence was 0% by clinical history, with a minimal follow-up of 1 year. With the same surgical technique Yu and Hamilton [562] had a positive history in 2 out 35 patients (6%) and a positive Minor test in 15% of patients. More recent results are less encouraging: Belly et al. [39] describe a 40% positive history in 45 patients and Linder [318] has over 60% of postparotidectomy positive Minor test.

While our study was in progress, Shemen [476] described the use of polytetrafluoroethylene (Gore-Tex®) after parotid gland resection to prevent the occurrence of retro-mandibular depression and Frey syndrome. The synthetic material was used in 9 patients and none had a post- operative history of gustatory sweating. No Minor test was done and the follow-up after the operation was not stated. Also recently, Laccourreye et al. [288] describe the use of Surgicel® as a barrier for preventing Frey syndrome – while detail numbers were not given, the incidence of Frey was high, with and without Surgicel®.

It is difficult to draw a firm conclusion concerning the appropriate surgical prevention technique for Frey syndrome during parotidectomy, because of the small number of patients and often lack of objective testing. It seems that some form of barrier between the parotid bed and the

125 overlaying skin might be useful. Also, probably deeper dissection or a SMAS flap might decrease the incidence of postparotidectomy Frey syndrome.

Prevention technique Evaluation Follow-up Incidence with Incidence p (χ2) prevention without prevention

Auriculotemporal nerve avulsion (Kidd History ? 20% 24% 0.7 [273])

Sternocleidomastoid flap (Kornblut [278]) History 1-6 years 43% 23% 0.07

Minor test 1-6 years 97% 94% 0.5

Sternocleidomastoid flap (Casler [85]) History 2 years 12.5% 47% < 0.01

Temporal fascia flap (Sultan [504]) History ? 0% ---

Thick skin flaps (Singleton [479]) History 8 weeks 3% 12.5% 0.02

Dermal-fat graft (Nosan [391]) History ? 0% ---

Dermal-fat graft (Harada [222] History 2-4 years 0% ---

SMAS dissection (Allison [7] ) History 1 year 1% 83% <0.01

Minor test 1 year 3% 87% <0.01

SMAS dissection (Casler [85]) History 2 years 0% 47% <0.01

SMAS dissection (Bonanno [51]) History 1 year 0% ---

SMAS dissection (Yu [562]) History 2.5 years 6% ---

Minor test 2.5 years 15% ---

SMAS dissection (Belli [39]) History ? 40% 57% 0.13

SMAS dissection (Linder [318]) Minor test 1 year 70%

Sternocleidomastoid flap (Sood [Sood, Minor test 1 year 18.2% 81.8% <0.05 1999 #933])

Table XVII: Effect of various surgical techniques for preventing Frey syndrome From Dulguerov et al. [136] without permission.

126 2.3. Wound complications Besides facial nerve paralysis and Frey syndrome, there is another set of complications, most of which occur after any type of surgery such as wound hematoma, seroma, flap necrosis, keloids or unaesthetic scarring, and local cutaneous anesthesia. We have called this group wound complications. Three other complications, included in this group, are specific to parotid surgery: retromandibular depression, salivary fistula, and auricular anesthesia.

Little has been written on these complications until the last 10 years. Even now, most of the literature is from the esthetic surgery literature on parotidectomy incisions and surgical correction/prevention of post-parotidectomy retromandibular depression.

2.3.1. Parotidectomy incisions

Apparently, the early incisions for parotid surgery were straight, vertical, or horizontal [165]. Often a horizontal limb of the incision was present over the cheek [48; 165] (Figure 30A, 30B). Guerrerosantos [210] and later Appiani and Delfino [11] reviewed parotidectomy incisions and credited Gutierrez, in 1903, to be the first to advocate an incision around and under the ear lobule. No clear reference to an article, even in Spanish, is given. A reference to a 1924 publication of Gutierrez is given [216], but this reference is incorrect, and therefore we are unable to confirm the claims of Guerrerosantos [210] and Appiani and Delfino [11].

Duval [149; 150] and Sistrunk [481] used only a neck incision, parallel to the mandible. Adson and Ott [3] extended this incision with a preauricular limb that hides in a preauricular crease. "This incision intersects the curved incision … that extends from 3 cm upward and posterior to the lower tip of the mastoid process and runs 2 cm below the lower border of the mandible, ending 4 cm anterior to the angle of the jaw" (Figure 30C).

A slight modification, essentially abandoning the retroauricular portion, is the incision used by Bailey [16] (Figure 30D). He called this incision a "modified Blair incision," although examination of Blair's textbook does reveal a straight incision. The incision recommended in Redon's book on parotid surgery does not have a curvature under the ear lobule [437] (Figure 30E). Bailey's and Redon's incisions are sometimes referenced, but they differ from the incision described in modern Head and Neck surgery textbooks [154; 262; 322; 477]. Bailey's incision retains the sharp angle between the preauricular and neck limbs and has been abandoned, probably because of skin flap necrosis, for the so-called "lazy S" incision [64; 220]. The "lazy S" incision (Figure 30H)

127 has the usual preauricular limb and a curved neck limb, which is supposed to follow a neck skin crease. These two limbs are joined by a progressively curved portion located behind the ear lobule. Its relatively broad anterior base provides for a good vascularization and renders this incision rather immune to skin necrosis. In reviewing the literature and the incisions drawn, it seems that the first to use a gently curved incision with a significant retroauricular portion were Brown et al., in 1950 [63].

Probably the first to use a face-lift incision for parotidectomy were Appiani, in 1967 [11], and Jost et al., in 1968 [264]. Appiani [11; 12], Hinderer [235], Hagan and Anderson [220], Guerrerosantos [210], Cohen [100], Allison and Rappaport [7] all mention the use of rhytidectomy incision for parotid surgery without any real evaluation of the esthetic results or the complications rate (Figure 30I). Ferreira et al. [170] were first to report complications with this approach such as keloids (14%), edema of the ear lobe (2%), salivary fistula (2%). Terris et al. [512] are first to compare the usual parotidectomy and the rhytidectomy incision in a retrospective, non- randomized, non-blinded trial. No significant difference in time or complications rate between the two approaches was apparent. There was 1 case each of hematoma, wound infection, and salivary fistula for an incidence of 6% in the face-lift incision group [512].

A rhytidectomy incision is probably adequate for most cases requiring a superficial parotidectomy, but might be inadequate for medially located tumors, for deep lobe tumors, and when a neck dissection is deemed necessary because of a malignancy [106; 220]. Extensions of the incision along or within the temporal hairline and a vertical extension close to the posterior midline in the neck have been proposed [100]. These appear appealing, but whether they can be used for large parotid resections and reconstructions [442] remains to be demonstrated.

128

A : Faure [1895] B : Blair [1912] C : Adson & Ott [1923]

D : Bailey [1941] E : Redon [1945] F : Kloop [1950]

H : Brown [1950] I : A ppiani [1967] G : Martin [1952] La zy “ S” and Jost [1968]

Figure 30: Parotidectomy incisions

129 2.3.2. Retromandibular post-parotidectomy depression

One of the first to write about preventing the post-parotidectomy depression was Jost [264] who used a superiorly based sternocleidomastoid flap, and Champy [91] who used a "retroauricular musculoperiosteal flap". Kornblut also used the SCM flap but was interested in preventing Frey syndrome and did not report about the cosmetic aspect [278]. Bugis et al. [70] used the sternocleidomastoid flap in 31 patients and reported 1 parotitis, 1 salivary fistula, and 1 hematoma. No formal evaluation of cosmesis is provided in the above-cited papers.

The SMAS flap technique is proposed by Rappaport and Allison [428] not only for Frey syndrome prevention but also for amelioration of post-parotidectomy defects: "the approach has resulted in greater acceptance of the tumor surgery by patients and less dissatisfaction postoperatively."

Nosan et al. [391] report on the use of dermal-fat graft taken from the abdomen and implanted in the parotid bed prior to closure. In 9 patients, the cosmetic aspect is said to be pleasing to all the patients, and the complications included 2 abdominal cellulitis, 1 facial cellulitis, 1 parotidectomy wound drainage attributed to fat liquefaction, and 1 parotidectomy wound hematoma. The authors state that no major long-term resorption of the implanted tissue is noticed.

The temporal fascia rotation flap [504] is also proposed for prevention of post-parotidectomy depression. In a group of 7 patients, the authors found that the contour was normal in 6, while a depression was present in 1 patient.

A simple method was recently described by Trotoux [521]: the use of a blood clot prepared from patient's one blood. The technique was used in 39 patients: in 8 cases a complete absorption of the clot was noted and in 2 other cases absorption was noticed with longer follow-up. Thus, the technique was inadequate in 25%. What will probably deter the use of this technique is the occurrence of wound hematoma requiring drainage and wound packing in 21 patients (54%).

Free dermal fat flaps micro-anastomosed to cervical vessels were reported on five patients having a radical parotidectomy, often extended to the mandible or the temporal bone, by Baker et al. [21]. The flap survival was discussed, rather than the esthetical appearance or complications.

While the above-cited publications deal with surgical techniques, without any formal evaluation, Arndt et al. [14] measured the postparotidectomy depression using a custom device.

130 No special surgical technique was used and the difference was appreciated in relation to the normal side: the depression was measured as 0.68 cm after superficial parotidectomy and 1.03 cm after total parotidectomy.

2.3.3. Salivary fistula and post-parotidectomy wound collections

In parotid surgery, the dissection always requires sectioning through parotid gland tissue. Therefore, some local accumulation of saliva under the skin is common. Because the treatment is often expectation and local compression, it is usual to try avoiding drainage of the collected fluid through the skin [336]. When the collection is limited and no external drainage is performed, we consider this collection as a seroma, although some of the fluid collected is probably of salivary origin. When drainage is performed, the removed fluid allows classification of the complication as hematoma (red and thick fluid), infection (turbid tan color fluid, sometimes thick and obviously purulent), or as salivary fistula (clear fluid). The definition of postparotidectomy salivary fistula used in the sole publication on the subject [542] was: 1) drainage from the incision or drain site that is clear, increasing with mastication, and without signs of infection; 2) initially cloudy or purulent discharge which clears after treatment to continue as clear fluid.

A 14% incidence of postparotidectomy fistula was found by Wax and Tarshis [542]. These fistulas were more frequent when parotidectomy was performed for an inflammatory condition (45%), as compared to benign (8%) or malignant (18%) tumors. Wound infections developed in 10% of patients. The management remains problematic since some of these patients had dressing changes for up to 35 days [542]. No other predisposing factors have been identified. The role of Stenson's duct ligature during parotidectomy to prevent salivary fistula, as suggested by Kun et al. [283], remains to be demonstrated.

Treatment options of salivary fistula have been most often studied in the context of post- traumatic etiologies [10; 406]. They can be divided in two major groups: 1) diversion of parotid secretion into the mouth (reconstruction of Stenson's duct by primary suture, interposition grafts, or creation of a controlled internal fistula); 2) depression of parotid secretion (Stenson's duct ligature, tympanic neurectomy, no oral intake, expectation for spontaneous closure, compression dressings, antisialogogues, radiotherapy) [10; 406]. Radiotherapy and tympanic neurectomy have been abandoned, and Stenson's duct ligature has little role in the treatment of postparotidectomy salivary fistula. The respective role of conservative management and antisialogogues is difficult to evaluate from the literature.

131 A promising new therapy is the somatostatin analog octreotide that has been described recently in a two cases [298; 487]. Somatostatin has been shown to decrease salivary secretion in man [321] and has been used to reduce fistula and other complications of pancreatectomy [172].

Recently, Depondt et al. [124] assessed the role of fibrin glue in parotidectomy wounds in a retrospective non-randomized, non-blinded study (Table XVIII). A similar influence of fibrin glue has been shown on postoperative hematoma after face-lift procedures [331] and on seroma after neck dissections in a rat model [331].

Controls Fibrin glue p Total patients 34 34 Hematoma major 5 0 0.02 Hematoma minor 1 2 NS Salivary fistula 1 2 NS Flap necrosis 1 0 NS

Table XVIII: Effect of fibrin glue on postparotidectomy wound complication Data from Depondt et al. [124]

2.3.4. Skin anesthesia

The sensory innervation of the pinna and external auditory canal is derived from the cervical plexus and cranial nerves V, VII, IX, and X. Branches of cranial nerves IX and X join to form Arnold's nerve, which enters the temporal bone along the jugular vein, to gain access to the middle ear [468]. According to Bruesch [67], Arnold's nerve joins the facial nerve after passing through the stapedius muscle, close to the takeoff of the chorda tympani [177]. After receiving fibers from the facial nerve, the "inferior branch of Arnold's nerve [468] exits the temporal bone through the temporomastoid suture [67] or the stylomastoid foramen [177]. This sensory nerve is supposed to innervate the posterior aspect of the external ear canal and concha, as well as a various amounts of the posterior aspect of the skin of the pinna [60].

132 The anterior aspect of the external auditory canal, the tragus, and a portion of the anterior aspect of the pinna receives sensory innervation from the auriculotemporal temporal nerve (see § 2.2) [111; 532]. The remaining, actually the majority of the anterior aspect of the pinna, receives sensory innervation from the great auricular nerve, a sensory branch of the cervical plexus (Figure 31). The separating line between the innervation from the trigeminal nerve and the cervical branches was extensively studied by H. Cushing [111]. Contrary to other systems where a significant overlap of sensory innervation is present, Cushing found little overlap between the trigeminal and cervical sensory distribution [111]. This might explain the persistence of a significant skin anesthesia after parotidectomy.

Figure 31: Distribution of the facial innervation by the trigeminal nerve From Cushing [111] without permission. Shaded area tactile innervation and dotted area pressure innervation.

Little has been written about postparotidectomy skin anesthesia, and even less on the relative contribution of the great auricular and auriculotemporal nerves in this deficit. Nevertheless, the consequences could be serious, since extensive thermal burns have been described [62; 162; 309]. While the cutaneous anesthesia is usually attributed to sectioning the great auricular nerve, the contribution to postparotidectomy skin anesthesia by the cutaneous branches of the auriculotemporal nerve has not been studied. Whether cutaneous branches of the auriculotemporal nerve are routinely cut during the dissection of the posterior aspect of the parotid gland (see §

133 1.4.3.) remains to be investigated, since it passes in the plane between the parotid gland and the external auditory canal.

The greater auricular nerve is easily found on the superficial surface of the sternocleidomastoid muscle, where it pierces the superficial layer of the deep cervical fascia (see § 1.2.1). Kinney and Katrana [352] describe the location of the great auricular nerve at the midbelly of sternocleidomastoid muscle, 6.5 cm caudal to the inferior aspect of the bony external auditory canal.

In 1955, Kidd already mentions attempting to preserve the great auricular nerve, although the results are not detailed [273]. In 1982, Hawe and Bell [225] mention trying to preserve the posterior branch of the great auricular nerve in anteriorly or superiorly placed parotid tumors with apparent success in 8 out of 49 cases (16%). Debets and Munting state that the great auricular nerve "was saved in 65% of patients, thereby avoiding loss of sensation in the ear lobe" [119].

Brown and Ord [64], in a non-randomized, non-blinded study, compared the two points touch discrimination in 12 patients at 3, 6, and 12 months after parotidectomy. In the group with preservation of the posterior branch of the greater auricular nerve, the anesthesia surface was less extensive. In both groups, skin anesthesia was present, the anesthetized surface decreasing with time, although some skin anesthesia persisted at 12 months.

Christensen and Jacobsen [95] studied the preservation of the posterior branch of the greater auricular nerve in a blinded fashion using a questionnaire and routine technique to examine temperature, touch, and pain sensation. In 50% of patients with the nerve cut, the deficit was judged to be important. While the results in the nerve-preservation group were significantly better, a considerable percentage of patients (30-40%) still had sensory deficits, even at the level of the ear.

134 2.4. Recurrence It is evident that the fate of the patient with mixed tumor depends on the surgeon who operates upon him for the first time. R.W. Utendorfer [527]

It is often said that the most important complication in tumor surgery is recurrence. While the factors involved in the recurrence of parotid cancers are of prime importance, these are not discussed here, and this section essentially discusses the recurrence of benign parotid tumors. Because pleomorphic adenomas are by far the most common parotid tumor, representing close to 60% of operated parotid lesions (see § 1.3.), an important part of the history of parotidectomy is related to the evolution of the understanding of this tumor, and the best way to treat it (see § 1.1.). For other benign tumors, recurrence rates have been either null [152; 221] or very low [396].

As early as the 1930s, McFarland had pinpointed the controversial issues on the subject of pleomorphic adenomas:

1) "Histological variations among mixed tumors have no bearing upon prognosis" [347].

2) "Mixed tumors are inherently benign, but commonly recur after excision, and if frequently disturbed become locally destructive and invasive giving metastasis" [347].

3) "Malignant change … in mixed tumors, must be very rare and its occurrence difficult to prove" [347].

4) "As intervals of 10, 20 and even 30 years may elapse between the operative removal of a mixed tumor and its recurrence, caution should be exercised in declaring any case to be cured" [347].

5) "It is not advisable to operate upon the tumors when small, as the smaller tumors are the more apt to recur" [347].

6) "Irradiation has not yet been shown to be of any benefit" [347].

2.4.1. Histology and recurrence

Saksela et al. [454] examined the cellularity, cell atypia, mitotic activity, the "encapsulation," and the presence of satellite or multiple nodules of pleomorphic adenomas. None of cytological criteria studied had a correlation with recurrence. The number of recurrences appears higher in tumors with satellite nodules, although the specific classification of this criterion is not given, and

135 no statistical tests performed. While finding similar results, McGregor et al. [349] found that "nodular" and "myxoid" pleomorphic adenoma tend to recur more frequently, a fact they attributed to increased friability and probably intraoperative spillage. Similar statements can be found elsewhere [168].

Although McFarland's statement has apparently stood the test of time, pleomorphic adenoma with a nodular macroscopic appearance should be resected more generously. Since the myxoid histology cannot be identified preoperatively, emphasis should be placed on a intraoperative atraumatic handling of all pleomorphic adenoma.

2.4.2. Initial surgery and recurrence

In general, surgery less than superficial parotidectomy is condemned in modern Head and Neck textbooks [154; 262; 477]. However, such limited surgery has enjoyed support in few publications [13; 104; 200; 224; 232; 248; 343; 493]. Figure 32 shows an average incidence of recurrence after three parotid operations (enucleation, superficial parotidectomy, total parotidectomy) in the publications listed in Appendix I.

The factors favoring recurrences were listed by Conley and Clairmont [107] and confirmed by others:

1) Seeding of the tumors during the initial operation [296; 330; 371; 522];

2) Enucleation, missing small projecting lobules [296; 330; 371; 522];

3) Rare multicentric tumors;

4) Tumors in intimate contact with the facial nerve, when superficial parotidectomy becomes comparable to a peri-capsular dissection. Fortunately in the pseudopodia-like growth of pleomorphic adenoma [414], capsular perforation is more frequent in the lateral aspect of the gland rather than in the deep aspect, against facial nerve branches [305] ;

5) Large tumors;

6) Deep lobe tumors [522] or, even worse, tumors with a deep lobe extension that was not recognized [330];

7) Tumors located in the superior edge of the gland where they can extend under the zygoma and a resection with a cuff of normal tissue is difficult to achieve;

136 8) Drain site implantation [330; 414];

9) Peroral parapharyngeal space approach [330].

35%

30%

25%

20% %

15% 14.38%

10%

8.20%

5% 6.70%

0% Enucleation Superficial parotidectomy Total parotidectomy

Figure 32: Average and standard deviation of postparotidectomy recurrence Enucleation is from [13; 35; 96; 117; 151; 194; 200; 205; 224; 245; 253; 301; 329; 343; 365; 389; 454; 480; 493; 528; 530], superficial parotidectomy [166; 194; 200; 224; 225; 245; 301; 329; 365; 389; 454; 480; 544; 545; 557; 560], total parotidectomy [6; 81; 104; 166; 200; 211; 273; 288; 314; 330; 336; 389; 456; 480; 482; 527; 544].

While inadequate excision of pleomorphic adenoma has been found to be associated with more frequent recurrences [152; 371; 522], its exact incidence is difficult to evaluate. In a 1988 survey conducted in the United Kingdom by Gunn and Parrott [213], incomplete excision of benign tumors was found in 33% of cases using enucleation and 12% of cases using superficial parotidectomy. Buchman et al. [69], after reviewing 76 cases of pleomorphic adenoma in which 17 cases (22%) had either a tumor spill or inadequate resection, concluded that positive margins were associated with recurrence, while tumor spill was not. Without comparing these two types of inadequate excision, several studies have found more recurrence after tumor spillage [97; 107; 296; 330; 522]. In the series of Grage et al. [205], Maynard [340], and Huber et al. [244], all recurrences were in patients with capsule rupture during the initial parotidectomy.

Following Redon, French surgeons often perform a total parotidectomy for the initial treatment of pleomorphic adenoma [94; 211; 270; 288; 520; 522]. Cannoni et al. [81] and Guerrier et al. [211] have the impression that the chances of recurrence are proportional to the salivary gland

137 tissue remaining in the parotid bed. While the results in Figure 32 do not seem support this point of view, the role of routine total parotidectomy in young patients might deserve further studies.

2.4.3. Age at initial surgery and recurrence

The influence of the age at initial parotidectomy on subsequent recurrence was specifically examined by McGregor et al. [349]. They found 75-100% recurrence when the operation was performed before the age of 20 and 10-20% recurrence in older patients [349]. As nicely put by Maran et al. [330], "the question that arises is whether the incidence of recurrence is higher if one lives longer after treatment as in the case of those having tumors in the younger-age group or is it a different disease with a different histologic behavior." Others have also found increased recurrence rates in younger patients [166; 167; 211; 296] and, therefore, some [211; 296] favor near total parotidectomy as the initial procedure of choice for pleomorphic adenoma in young patients.

2.4.4. About small tumors and capsules

McFarland's conclusion from the 1930's, that small mixed tumors recur more often than large ones and therefore that small tumors should not be operated on [347], was finally explained and refuted by Patey and Thackray [414]. As discussed earlier, they examined resected mixed tumors with serial sectioning and found that: 1) the capsule surrounding the tumors was found to vary greatly in thickness not only from case to case, but also in different areas of the same tumor; 2) the capsule was found sometimes to be incomplete with tumor adjacent to normal gland tissue; 3) mixed tumors were found to be nodular and to grow by polypoid extensions through the capsule; 4) these polypoid extensions were more common in small tumors; 5) no multifocal mixed tumors were found in 37 specimens [414]. After recurrence, mixed tumors were often multicentric in the entire operated field and in particular along the previous scar [414]. These conclusions have been confirmed by Gunnel [214], Kleinsasser [275], Naeim et al. [372], Danovan and Conley [113], Lam et al. [299], and Lawson [305].

2.4.5. Malignant degeneration of pleomorphic adenoma

Malignant mixed tumors have been described, not infrequently, after a long history of pleomorphic adenoma [168; 211; 221; 296; 330; 389; 420; 454]. While this is usually attributed to a clonal malignant degeneration, the possibility that a recurrent pleomorphic adenoma represents an initially misdiagnosed malignant mixed tumor is to be kept in mind [97; 107; 160; 168; 301].

138 2.4.6. Surgery for recurrences and its complications

Re-exploration of the parotid region in a patient whose entire superficial lobe has been removed and whose entire facial nerve has been dissected in my hands has been a disaster, if the primary goal is identification, preservation and re-dissection of each branch of the facial nerve from the stylomastoid foramen to its muscle innervation. In conclusion, the solution of the problem requires a very understanding patient, and a surgeon who is ready for a challenge; not only to overcome the technical aspects of the problem but also to overcome the medico-legal aspects as well. Cocke, 1978 [97]

Overall, higher complication rates are observed after revision parotidectomy [81; 105; 107; 211; 221; 330; 396].

Facial nerve damage is more frequent in surgery for recurrences [81; 97; 152; 211; 221; 330; 371; 389; 396; 559], because of seeding [152; 221], incorporation of facial nerve branches in recurrent tumors [152; 330; 371; 389; 559] [396], resection of facial nerve branches [221; 396; 559], more traction on the nerves during surgery [152], preoperative facial nerve weakness [107; 221], blurring of tissue planes [221; 371], scarring [81; 97; 107; 211; 221; 330; 371; 389; 396; 559], and distortion of anatomical landmarks [81; 221].

The extent of surgery for recurrent pleomorphic adenoma is very controversial. The publication of Work et al. [559] is one of the first to blatantly state that facial nerve branches, if embedded or close to a recurrent tumor, should be sacrificed. Such an attitude is also favored by others [38; 107; 330; 389]. Other publications, which recommend routine identification and superficial parotidectomy as a minimal procedure for parotid tumors at their initial presentation, tend however, to favor deviations from this rule in recurrent cases to avoid an increased incidence of facial paralysis [168; 420]. Nodule excision, possibly after identification of a neighboring branch, is then recommended [105; 107; 168; 420]. Conley [105; 107] has favored this approach in cases presenting with a unifocal recurrent pleomorphic adenoma. However, Niparko et al. [389] found a poorer control of the disease and a higher incidence of facial injury with excision without formal facial nerve dissection.

Even when surgery is not limited to nodule resection, the extent of parotidectomy in recurrent pleomorphic adenoma is controversial. Some favor total parotidectomy with dissection

139 of the deep lobe [330; 371; 389], and others have questioned this in cases of superficially located recurrences [371; 420]. Since the majority of recurrences is in the scar or lateral to the facial nerve [371], the topic remains wide open.

The position of the recurrence, as a factor influencing the timing of the reoperation, was discussed by Fee et al. [168]: peripheral nodules can be observed for a much longer time than retromandibular ones.

2.4.7. Re-recurrence of pleomorphic adenomas

Higher recurrence rates are found after surgery for recurrent pleomorphic adenomas [38; 78; 107; 152; 211; 281; 296; 330; 371; 389; 440].

Multiple recurrence foci are the rule rather than the exception [81; 107; 152; 211; 221; 296; 330; 389; 420; 559], therefore, en-bloc removal of the tumor, scar, and possibly facial nerve branches is advised [97; 559]. According to Maran et al. [330], implantation recurrence, which is multicentric, is due to rupture of the capsule and spillage during the initial operation, while enucleation results in often unique regrowth of the pleomorphic adenoma. Cannoni et al. found that multicentric recurrences are noticed early, within 3 years of the initial parotidectomy, while recurrences after 10 years were unifocal [81].

2.4.8. Radiation of pleomorphic adenomas

In view of the high recurrence rates of mixed tumors in earlier series, radiation therapy has been proposed as adjuvant therapy since the 1920's [74], either as external beam radiation [5; 13; 224; 247; 343; 407; 528], curietherapy with radium implants [74; 343; 528], or an association of both [5; 247]. Although the available results do not show a reduction of recurrence rates by radiation [454; 522; 528; 559] or a minimal effect [77; 211; 389; 420], proponents of radiation have continued until the 1980's [13; 25; 194; 224; 343; 528].

The role of radiation therapy after recurrence of pleomorphic adenoma is still debated. Some authors still advocate radiation in recurrent cases, preferring to leave microscopic disease against the facial nerve rather than sacrificing the nerve [456], while others conclude against it [69; 94; 539].

Even if postoperative radiation is believed to be efficacious, this treatment modality should, probably, be avoided in young patients because of the increased incidence of radiation induced

140 parotid malignancies [118] and the possibility that radiation might promote a potential malignant transformation of pleomorphic adenoma [117].

3. OBJECTIVES

In broad terms, the goals of this work are:

1) to review of the incidence, the causative factors, the treatment and prevention strategies of parotidectomy complications;

2) to develop new techniques to study objectively these complications;

3) to find new ways to decrease these complications.

More specifically, we focused on the:

1) Development of an objective test for the evaluation of facial nerve function;

2) Evaluation of the role of continuous EMG-based intraoperative facial monitoring in the prevention of postparotidectomy facial paralysis;

3) Study of the facial nerve branches most at risk during parotid surgery;

4) Study of the delay of recovery of facial nerve paralysis;

5) Development of objective test for facial sweating and their use for the measurement of Frey syndrome;

6) Role of different surgical techniques for the prevention of Frey syndrome;

7) Role of botulinum toxin in the treatment of Frey syndrome;

8) Evaluation of the incidence of the different wound complications after parotidectomy.

141 4. PATIENTS AND METHODS

This is a prospective, non-randomized study of patients undergoing parotidectomy. The focus is the assessment of the complications of the operation. Several new surgical techniques and devices were used to decrease the incidence of these complications. In order to evaluate these complications, new objective testing methods were developed.

4.1. Population From April 1994 to April 1998, all patients undergoing parotid surgery at the Clinic of Otolaryngology Head and Neck Surgery of the University of Geneva were offered to participate in this prospective non-randomized trial. The beginning of this study antedates the creation of an Ethics Committee in our department.

The preoperative evaluation included thorough history and clinical examination. Some patients underwent a fine-needle aspiration or radiological studies, such as CT scan, MRI, and ultrasonography. Most of these preoperative studies were performed before the patient was referred to our clinic or requested on an individual basis, without specific guidelines. The role of the various preoperative tests in establishing the diagnosis of parotid masses and the indication to parotid surgery are not part of this study.

Besides the usual preoperative explanation and consent, all patients underwent a preoperative evaluation of facial nerve function (videomimicography - see § 4.3. below). Facial motor function was also graded according to the House-Brackmann scale.

4.2. Surgery In all patients, parotidectomy was performed using standard surgical techniques (see § 1.4.) with several improvements. An intraoperative facial nerve monitoring was used in all patients. In the first 35 patients, a custom mechanical transducer was used, while, since April 1996, a commercial EMG-based nerve-monitoring device was employed.

The mechanical transducer was placed in the mouth, between the cheek and the teeth. The apparatus transduced the mechanical pressure generated by the contraction of the buccinator, and possibly other midfacial muscles into an audible alarm. This system provided feedback on stimulation of the buccal branches or of the entire facial nerve.

142 The EMG-based facial nerve monitor (Neurosign 100 - Magstim Company Inc., Spring Gardens, United Kingdom) consists of a differential EMG recording on 2 channels. Because of the importance of eye and mouth closure in facial motor function (see § 2.1.3.), the orbicularis oculi and oris were monitored (see Figure 10). Qualitative auditory and semi-quantitative visual feedback of the stimulated branches or nerve is provided.

During the surgical procedure, facial nerve handling was as atraumatic as possible. The number of facial nerve stimulations was kept to the minimum necessary. When the mechanical transducer was used, monopolar commercial stimulators were used with the lowest electrical current setting that generated a response. With the EMG-based nerve monitor, a bipolar electrical stimulating probe is provided and used at current levels less than 0.05 mA.

Facial branches that were intentionally or accidentally sacrificed were noted on a ad hoc drawing. In the analysis, section of a facial nerve branch (usually small peripheral branches), in the context of a parotidectomy for a benign tumor, was always classified as accidental. Deliberate sacrifice of the facial nerve or its branches was not an exclusion criterion. Therefore, all patients operated on were included in the data analysis.

The parotidectomy operation was classified in superficial, total and radical parotidectomy (see § 1.4.2.). While enucleation was never performed, some procedures classified as superficial parotidectomy could have been better named lateral superficial parotidectomy [394]. Also, some procedures classified as total parotidectomies should be named near-total parotidectomy [394].

During the study period, several implants were placed, before skin closure, as a mechanical barrier, in order to prevent the occurrence of a postoperative Frey syndrome (see § 2.2.8.). The choice of the individual implant was left to the individual surgeon, nevertheless in general, the use of different materials was chronological: 1) a mesh made of Polyglactin 910 and Polydioxanon (Ethisorb®, Ethicon, Spreitenbach, Switzerland); 2) lyophilized dura (Lyodura®, B. Braun Melsungen AG, Melsungen, Germany); 3) expanded polytetrafluoroethylene (Gore-Tex® - SAM Facial Implant, W.L. Gore & Associates, Inc., Flagstaff, USA). Some patients did not have any implant placed.

Intraoperative variables examined include the type of parotidectomy, the type of neck dissection, and the duration of the procedure. Other technical details noted include the technique of facial nerve monitoring, the number and intensity of facial nerve electrical stimulations, whether

143 facial nerve branches were sectioned, the use of subcutaneous barrier to prevent Frey syndrome, and whether the great auricular nerve is preserved or not.

The histopathology of the parotid lesions removed was classified according the WHO classification [478]. The size of the lesion was estimated by the pathologist by measuring the macroscopic lesion with the parotidectomy specimen.

Categorical data were compared with the exact Fisher test, while relationship between continuous variables were analyzed with the Pearson correlation test using the statistical algorithms of the SPSS 7.5 software (SPSS Inc., Chicago, IL, USA).

4.3. Post-operative evaluation The occurrence of postoperative wound complications was noticed daily during the hospital stay and the patients were asked about and examined for at each postoperative visit. Skin anesthesia was evaluated by history. Postoperative hematoma was considered to be present when the wound was open and bright red fluid or material was expressed. Postoperative fistula was considered to be present when clear fluid was expressed from the wound. The duration of wound care by dressing changes or rinsing was noted. If a collection of fluid was present under a closed wound and no drainage was required, this complication was called a seroma.

Because previous evaluation techniques of motor facial nerve function have been subjective and controversial (see § 2.2.), a new objective facial nerve evaluation method was developed and called videomimicography (see § 4.4, below). Videomimicography (VMG) was performed for each patient preoperatively and postoperatively at 1 week and approximately 6 months after the surgery. For patients with abnormal facial nerve function at 1 week, VMG was repeated at monthly intervals until normalization. At each VMG session, the facial motor function was also graded according to the House-Brackmann scale [243].

Similarly, because the often-used Minor test is cumbersome and qualitative, a new set of objective quantitative test for the evaluation of Frey syndrome was developed (see § 4.5, below). The patients were seen at least one year after parotidectomy to assess the presence of Frey syndrome. The average follow-up was 24 ± 13 months. Patients presenting an objective syndrome Frey syndrome were offered treatment with botulinum toxin (see § 4.6, below).

144

Figure 33: Videomimicography set up.

145 4.4. Videomimicography

4.4.1. Description

The objective facial nerve monitoring test developed for this study has been named videomimicography (VMG). The test is filmed on a videotape for analysis. During VMG, the subject is asked to produce 5 stereotyped facial movements (mimics). After the video film is digitized, graphical measurements of facial movements are made.

During VMG, the subject is sitting comfortably on a custom chair (Figure 33). The chair has a headrest, providing for a head support and minimizing head movements, both in the anteroposterior and in lateral directions. In addition, an adjustable frame allowed for the presence, just superior to the forehead, of a 10 cm scale which was used for the calibration and the measurement of actual facial distances.

Eleven facial landmarks were placed with a colored (blue) eyeliner pen (Figure 34). The landmarks used were similar, but not identical, to the ones proposed by Burres [72]. The landmarks used were:

1) Nasal (Na): placed in the midline, on the nasal bone, slightly below the nasion [186]. This point was found by Frey et al. [186] to correspond to an immobile point during facial movements and was, therefore, used as a reference point for the measurement of other facial landmarks;

2) Frontal (F): placed about 2 cm above eyebrow, above the pupil;

3) Infra-orbital (Io): placed at the level of the skin overlaying the orbital rim on a vertical line passing from the pupil;

4) Alar (A): placed a few millimeters lateral to the inferior edge of the nasal alae;

5) Mouth (M): placed a few millimeters lateral to the corner of the mouth;

6) Lip – superior (Ls) : placed in the midline, in the deepest point of the philtrum. It is usually a few millimeters, above the cupid bow line;

7) Lip – inferior (Li) : placed in the midline, in the deepest point of the chin. It is usually about 10 millimeters from the inferior vermilion border, at the midline.

Landmarks F, Io, A and M are bilateral, while Na, Ls and Li are at the midline. The landmarks were placed somewhat arbitrary without precise measurements of their exact location.

146 During the course of these measurements, we found that in case of severe paralysis (HB>3), maximal smiling efforts resulted in the displacement of landmarks Ls and Li towards the healthy side. This most probably resulted from the unilateral and unopposed contraction of the non- paralyzed facial muscles, and generated a “false” increase of the distance and area measurements on the paralyzed hemiface [534]. To deal with this error, Ls and Li were adjusted using the facial midline horizontal coordinate, i.e. the Na horizontal coordinate (Figure 34).

Figure 34: Videomimicography facial landmarks.

The 5 mimics chosen are routinely used in facial nerve testing (see § 2.1.2) and include: eyebrow rising, eyelid closure, nose lifting, lip puckering, and smiling. Prior to actual recording the requested movement was explained to the subject and a few trials performed. Beside nose lifting, which was not always possible to achieve, subjects had no trouble producing the mimic. For each mimic the subject is actively stimulated by verbal commands to produce the maximal possible movement and to keep this position for a few seconds.

147 For all of the subjects, a digital video camera was used (Panasonic, AG-EZ1), and the film recorded on a mini-DV digital videocassette. The entire video film was then replayed on a digital VCR (Panasonic, DV10000). For each subject 3 rest frames and the 3 frames for each mimic were selected and then fed to a PC-compatible computer under software control (DV Studio, Panasonic).

Figure 35: Adjustment of landmarks Ls and Li for smiling movement Ls → Ls’, Li → Li’. Ls’ has the X value of point Na and Y value of Ls. Li’ has the X value of Na and Y value of Li.

Measurements of the distances and areas were performed with the Osiris public domain image analysis software. A custom modification was used which allowed sending the coordinates of the marked point to a spreadsheet file, by clicking on them. This modification also performs the special calculation in terms of distances and areas, as well as their calibration in cm. The modified version of the Osiris software can be requested from the Division of Medical Information of the University Hospital of Geneva (www.expasy.ch).

From placing the facial landmarks to finish of recording, the total procedure took about 45 seconds for one repetition of the five facial movements in each subject. When 3 repetitions were used the total time of the test session was less than 2 minutes.

148 4.4.2. Subjects

Five normal control subjects (2 males and 3 females), with an average age of 41.2 years and no prior history of facial paralysis, underwent a videomimicography recording.

Twenty-nine patients with facial paralysis were offered to participate in this study. There were 16 males and 13 females, aged from 12 to 71 years (mean age: 39.7 ± 12.3 years). The etiology of facial paralysis was parotidectomy in 18, mastoidectomy in 5, temporal bone fracture in 2, Bell’s palsy in 2, and cholesteatoma in 2. None of these patients had received operations to rehabilitate their facial paralysis. The distribution of HB grades in our patient population was the following: HB1 – 5, HB2 – 9, HB3 – 8, HB4 – 5, HB5 – 2, and HB6 – 5.

4.4.3. Normative measures

In order to obtain the relevant measures for the objective measurement of facial motor function (see § 2.1.2.), various parameters were initially measured. For each movement, 10 distances and 5 areas were evaluated on the two half-faces.

The distances studied were (Figure 36): ALs (distance A-Ls), AM (distance A-M), FIo (distance F-Io), FNa (distance F-Na), IoA (distance Io-A), IoM (distance Io-M), MLi (distance M- Li), MLs (distance M-Ls), NaA (distance Na-A), and NaIo (distance Na-Io). The 5 areas (Figure

37) evaluated were: EYE area (ΣEYE), LATERAL area (ΣLATERAL), PARANASAL area (ΣPARANASAL),

UPPERLIP area (ΣUPPERLIP), and MOUTH area (ΣMOUTH). The percent change (ΔX) of each of the above 15 variables between rest and maximal movements were computed using the following formula:

Xmovement − Xrest X =Δ × 001 (with X representing a given measure) Xrest

For each facial movement the "best measure" was defined as the measure exhibiting the largest change relative to the rest value (high ΔX), while the standard variation of the measure was small.

In order to obtain an overall facial nerve function number, two global values were introduced: a VMG score (VMGs) and a VMG index (VMGi) [145]. These global values factor the importance of the eye and mouth sphincters in facial motor function: the "best eye measure" was

149 multiplied by 4, while the "best smiling measure" and the "best lip puckering measure" were each multiplied by 2. The exact formula for the VMGs is:

()ΔΣ eyein closure 4 ()ΔΣ+× foreheadin lifting (ΔΣ+ nosein wrinkling) (ΔΣ+ lipin puckering 2 )(ΔΣ+× smilingin × 2 ) VMGs ≡ EYE EYE PERANASAL UPPERLIP MOUTH 10

Figure 36: Videomimicography distance measurements. Ten distances were evaluated on the two half-faces: ALs (distance A-Ls), AM (distance A-M), FIo (distance F- Io), FNa (distance F-Na), IoA (distance Io-A), IoM (distance Io-M), MLi (distance M-Li), MLs (distance M-Ls), NaA (distance Na-A), and NaIo (distance Na-Io).

The VMGi is the ratio of VMGs of the paralyzed to the normal side:

VMGs paralyzed side VMGi ≡ × 001 VMGs normal side

150

Figure 37: Videomimicography surface measurements. Five areas were evaluated: EYE (area F-Na-Io), LATERAL (area A-M-Io), PARANASAL (area A-Na-Io), UPPERLIP (area A-Ls-M), and MOUTH (area Ls-Li-M).

4.4.4. Statistical analysis.

The coefficient of variation was computed using a standard formula: standard deviation/mean. The reliability of VMG was assessed by evaluating same day and day-to-day variability: the recordings for each normal subject were repeated on 4 different days (days 0, 1, 7, and 8), with 3 repetitions each day. Side-to-side, day-to-day, and retest variability (intrasubject variability), as well as intersubject variability were assessed by analysis of variances (ANOVA). Correlation between the HB grade and the individual measures and indexes was assessed with the Pearson correlation coefficient. ANOVA was also used to assess the factors involved in the total variability exhibited by the patients. The software used for the statistical tests was SPSS for Windows (version 9.0).

151 4.5. Facial gustatory sweating evaluation The presence of Frey syndrome was evaluated with two methods specifically developed for this study: the iodine-sublimated paper histogram (ISPH) method and the blotting paper technique.

Both the blotting paper and the iodine-sublimated paper were cut according to a custom stencil (Figure 38) that was adapted to the rather complicated topology of the lateral facial area. In both methods, the stencil was applied on both sides of the face during a gustatory stimulation, for a duration of 1 minute. The controlateral side was used as a control.

Gustatory stimulation was evoked by the suction of slice of lemon. The lemon slice was placed in the mouth and the patient asked to gently suck on the slice, without undue chewing.

Figure 38: Preformed stencil for Frey syndrome evaluation. The shape of the stencil resembles a fish with: a) short cut on the right to be aligned with the lip commissure, b) a large deep cut superiorly to fit around the pinna, and c) a narrow deep cut inferiorly. The narrow deep cut is located over the neck, below the mandibular angle, and was necessary to prevent undue paper folding during the testing. The cut stencils are symmetric and 2 stencils can be cut out of one sheet of paper in US letter or European A4 format.

152 4.5.1. Blotting paper technique

The blotting paper technique is a quantitative measure of the amount of sweat generated. It is simply the measure of the weight change in a blotting paper after the absorption of the sweat. A paper stencil is cut from commercial blotting paper, weighted, applied to the face during a gustatory stimulation, and weighted again. The weight change is taken to represent the amount of sweat absorbed.

The technique was calibrated by applying known quantities of 0.9% solution of sodium chloride on the stencils. The following amounts were used 2, 5, 10, 25, 50, 100, and 250 μl, delivered by commercial micropipettes. For each amount, the measure was repeated 3 times and the results reported as an average ± standard deviation.

Initially we noticed wide and erratic variations when weighting these stencils. The source of the problem was pinpointed to the distribution of the weight over the surface of usual commercial balances, and possible movements of the folded stencils. To correct this, of actual folding of the stencils was standardized and a special cut made to stabilize the folded paper and prevent its movements (Figure 39). With this technique, the margin of error of the method is around 2 mg (2 μl).

Figure 39: Standardized folding of the blotting paper stencil.

153 4.5.2. Iodine-sublimated paper histogram (ISPH)

The ISPH method [144] is a modification of the classical Minor test [360] (see § 2.2.5.). It is used to provide a quantitative evaluation of the amount of sweat produced as well as a topographic image of the sweat-producing area.

This method uses regular office paper sheets, which have been sublimated with iodine. One hundred office paper sheets are placed in a glass jar and exposed during 2 weeks to iodine vapors (1 gram of Iode resublimiert I-3380, Sigma AG, Deisenhofen, Germany). The iodine-sublimated paper takes on an amber color. Wetting this paper results in a localized color change, from amber to blue, similarly to color changes of the Minor iodine-starch mixture (Figure 40). As with the Minor test, this color change corresponds to the reduction of starch by water.

Figure 40: Iodine-sublimated paper stencil wetted with water.

154 The iodine-sublimated paper sheets were cut according to the predefined stencil, and applied to the face during a gustatory stimulation. After this, the stencil was scanned, and the digital image was subjected to a color histogram algorithm. We used a Scanjet II (Hewlett Packard) scanner in a 8 bit gray scale mode with a resolution of 70 dpi (28 points/cm or 784 points/cm2). The image analysis software used was Mathlab (The Math Works Inc., Natick, MA, USA). Since the stencils were scanned in an 8-bit mode, the available range of darkness is from 0 to 256 (X-axis, Figure 41 - lower left). The amber color of the stencil background was used as a threshold value. The histogram data darker (lower values) were than divided in 3 bins of equal width and of increasing darkness (Figure 41). The darkest bin corresponds to paper zones in which the starch is supposed to be totally reduced by the applied water (sweat) and only the surface of this darkest bin was taken into account to compute the wet surface (bin #1, Figure 41 - lower right).

The technique was calibrated by applying known quantities of 0.9% solution of sodium chloride on the stencils [144]. The following amounts were used 2, 5, 10, 25, 50, 100, and 250 μl, delivered by commercial micropipettes. For each amount, the measure was repeated 3 times and the results reported as an average ± standard deviation.

A correlation of the results obtained by ISPH and the blotting technique methods was made using the Pearson correlation test as implemented by the SPSS software.

4.6. Frey syndrome treatment with botulinum toxin Patients with clinical Frey syndrome, which was confirmed by the objective tests, were offered treatment by intradermal injection with botulinum toxin [143; 425]. Fifteen patients agreed to participate. We also recruited 5 patients who were either operated elsewhere or operated at our institution prior to the start of this prospective trial. One of these patients had bilateral parotidectomy and also bilateral Frey syndrome. Therefore, 16 hemifaces were used for injection with botulinum toxin.

Injection sites were 1 cm apart. Injection was made with a 30-gauge needle. Botulinum toxin at a concentration of 5 I.U. per 0.1 ml was used. At each injection site, about 0.1 ml (5 I.U.) of botulinum toxin was infiltrated. Statistical comparison of the measures taken before and after Botox injection was performed with the bilateral Student t-test using the statistical algorithms of the SPSS 7.5 software (SPSS Inc., Chicago, IL, USA).

155

Figure 41: ISPH data sheet of a patient with Frey syndrome TOP: Printout of the digitized stencil. The stencil resembles a fish several cuts: a) a small cut located anteriorly on the patient for the corner of mouth, b) a wide deep cut located superiorly for the ear lobe, and c) a narrow deep cut inferiorly located in the neck, below the angle of mandible, necessary to prevent undue folding during testing. LOWER-LEFT: The gray scale histogram with an X-axis corresponding to the degree of darkness with 0 being black and 256 being white. The threshold algorithm has assigned a value of 145 for the amber color of the stencil background, so everything above this value should be regarded as non-relevant. The remaining values (0 to 144) are divided in three histogram bins of equal width (in this case 48, because 144/3 = 48) of increasing darkness. The darkest bin is called bin #1, the middle one #2, and the less dark bin is # 3. LOWER-RIGHT: The percent of the total surface and the actual surfaces in cm2 for each histogram bin. Only the value of the darkest bin (#1) are taken to represent wet paper, and therefore are used in further calculations.

156 5. RESULTS

5.1. Videomimicography – the best measures Beside forehead and nose wrinkling, which were sometimes difficult to perform, subjects had no trouble in producing the requested facial movements. For every facial movement studied, the measures exhibiting highest changes relative to rest (ΔX) were always measures involving surfaces rather than distances (Table XIX). For eye closure and forehead lifting, the measure with the largest percent change relative to the rest value was ΣEYE, with -31.86 ± 8.54% and 12.73 ± 4.84% respectively (Figure 42 and 43). For nose wrinkling the best measure was ΣPARANASAL, with an average change of -28.08 ± 9.50% (Figure 44). For lip puckering the best measure was ΣUPPERLIP, with an average change of –22.89 ± 8.29% (Figure 45). For smiling the best measure was ΣMOUTH, with an average change of 63.48 ± 21.27% (Figure 46). For all of these five best measures, the coefficient of variation was smaller than 0.4.

EYE FOREHEAD NOSE LIP SMILING CLOSURE LIFTING WRINKLING PUCKERING MEASURE (mean ± SD) (mean ± SD) (mean ± SD) (mean ± SD) (mean ± SD)

ΣEYE -31.86 ± 8.54 12.73 ± 4.83 -21.36 ± 9.94 -2.30 ± 9.03 -9.99 ± 8.76 ΣLATERAL -16.32 ±16.01 0.58 ± 10.83 -4.48 ± 19.88 -18.22 ± 9.96 -8.63 ± 14.88 ΣPARANASAL -25.08 ±14.49 4.83 ± 12.19 -28.08 ± 9.49 0.72 ± 10.86 -16.60 ± 12.85 ΣUPPERLIP 5.49 ± 10.02 0.23 ± 8.96 13.22 ± 12.73 -22.89 ± 8.29 21.53 ± 20.71 ΣMOUTH -1.30 ± 8.87 -0.81 ± 12.36 0.67 ± 16.43 -13.88 ± 9.14 63.48 ± 21.27 Als 6.19 ± 9.55 -0.34 ± 6.58 10.26 ± 9.89 -11.43 ± 7.55 15.15 ± 9.03 AM 3.04 ± 7.03 0.26 ± 4.73 11.86 ± 9.07 -9.86 ± 5.88 3.07 ± 14.03 FIo -21.00 ± 7.80 10.23 ± 6.56 -16.72 ± 7.26 -0.01 ± 6.16 -10.82 ± 7.21 FNa -11.60 ± 3.88 5.95 ± 4.26 -3.89 ± 6.87 0.65 ± 4.19 -1.06 ± 3.54 IoA -4.91 ± 10.71 0.01 ± 6.31 -15.66 ± 11.33 3.08 ± 6.38 -11.08 ± 8.58 IoM 3.49 ± 4.87 0.19 ± 3.32 0.83 ± 6.03 -1.85 ± 4.77 -2.73 ± 7.15 MLi 3.73 ± 6.52 0.16 ± 7.06 1.04 ± 9.48 -15.71 ± 5.68 32.83 ± 9.69 MLs -1.93 ± 5.34 0.56 ± 6.56 -3.76 ± 6.47 -16.57 ± 7.15 15.49 ± 8.88 NaA -12.76 ± 8.34 7.26 ± 4.46 -22.31 ± 8.59 -1.48 ± 4.65 -8.06 ± 6.42 NaIo -20.25 ± 8.53 4.86 ± 5.56 -15.67 ± 6.99 -3.35 ± 5.46 -6.53 ± 5.43

Table XIX: Percent change of all measures for each movement in normal subjects. See text for abbreviations. Note that some movements, such as eye closure, nose wrinkling, and lip puckering, result, for most measures relative to rest, in decreases and therefore negative values. Favorable measures exhibit large change relative to the rest value, while the variation of the measures is small. The best measures for each movement are shaded. From Dulguerov et al.[145], without permission.

157

Na-Io Na-A M-Ls M-Li Io-M Io-A F-Na F-Io A-M A-Ls MEASURE MOUTH UPPERLIP PARANASAL LATERAL EYE

-50 -40 -30 -20 -10 0 10 20 PERCENT CHANGE

Figure 42: Percent change for eye closure in normal subjects

Na-Io Na-A M- Ls M- Li Io- M Io- A F-Na F-Io A-M A-Ls MEASURE MOUTH UPPERLIP PA RA NA SA L LATERAL EY E

-15 -10 -5 0 5 10 15 20 PERCENT CHANGE

Figure 43: Percent change for forehead lifting in normal subjects

158

Na-Io Na-A M- Ls M- Li Io- M Io- A F-Na F-Io A-M A-Ls MEASURE MOUTH UPPERLIP PA RA NA SA L LATERAL EY E

-50 -40 -30 -20 -10 0 10 20 30 PERCENT CHANGE

Figure 44: Percent change for nose wrinkling in normal subjects

Na-Io Na-A M- Ls M- Li Io- M Io- A F-Na F-Io A-M A-Ls MEASURE MOUTH UPPERLIP PA RA NA SA L LATERAL EY E

-40 -30 -20 -10 0 10 20 PERCENT CHANGE

Figure 45 Percent change for lip puckering in normal subjects

159 Na-Io Na-A M- Ls M- Li Io- M Io- A F-Na F-Io A-M A-Ls MEASURE MOUTH UPPERLIP PA RA NA SA L LATERAL EY E

-40 -20 0 20 40 60 80 100 PERCENT CHANGE

Figure 46: Percent change for smiling in normal subjects

5.2. Videomimicography – variability in normals Analysis of variance (ANOVA) revealed that, for most of the measures, the majority of the total variation resulted from intersubject variability, while side-to-side, day-to-day and retest

(intrasubject) variability were negligible (Table XX). For ΣEYE in eye closure and forehead lifting movements, ΣPARANASAL in nose wrinkling, ΣUPPERLIP in lip puckering and ΣMOUTH in smiling, ANOVA showed that intersubject variation was responsible for, respectively, 91%, 73%, 97%, 82%, and 90% of the total variability. Depending of the best measure studied, the percent of total variability due to side variation was between 0.2 and 2%, to same day retest variation was between 1 and 13%, to different days retest variation was between 1% and 12%. Only the intersubject variability reached statistical significance (p<0.05).

For intrasubject variability, the majority of high values were found in forehead lifting movements, with intermediate values in lip puckering, and low values in eye closure, nose wrinkling, and smiling.

160 Day Side Repetition Subject

Measure Movement p % p % p % p %

ΣEYE Eye closure 0.344 1% 0.153 7% 0.109 1% 0.000 91%

ΣEYE Forehead lifting 0.403 2% 0.296 12% 0.121 13% 0.000 73%

ΣPARANASAL Nose wrinkling 0.283 1% 0.754 1% 0.429 1% 0.000 97%

ΣUPPERLIP Lip puckering 0.703 0.4% 0.215 10% 0.200 7.6% 0.000 82%

ΣMOUTH Smiling 0.564 0.2% 0.300 5% 0.202 4.8% 0.000 90%

Table XX: ANOVA of side, day, subject and repeat variable P = p value; “%” represent the percentage of the square for a variable in the total square. The sum of “%” across a given row is 100.

5.2. Videomimicography – correlation with the

House-Brackmann scale With increasing degree of facial paralysis, the magnitude (in absolute value) of all of the best measures decreased from their maximal value (Table XXI). ΣEYE in eye closure decreased from

32% to 1%; ΣEYE in forehead lifting closure decreased from 13% to 1%; ΣPARANASAL in nose wrinkling decreased from 28% to 4%; ΣUPPERLIP in lip puckering decreased from 23% to 4%; ΣMOUTH in smiling decreased from 63% to 6%. Percent changes of these best measures in total facial paralysis (HB = 6) were close to 0.

Similar trends were found with the global indexes. VMGs decreased from 34.10% to 4.94%, and VMGi decreased from 103.02% to 18.03%. All of the best measures (Figure 47) had high and statistically significant correlation with the HB grade (Table XXI). The global values, VMGs and VMGi (Figure 48) also demonstrated a correlation with the HB grade (Table 2). In particular, the Pearson correlation coefficient for VMGi was excellent (-0.94).

ANOVA revealed that for all of these measures, side-to-side and retest variability were very small and negligible. Most of the overall variability was due to the HB grade, which was the only variable to reach statistical significance (p<0.001) (Table XXII).

161

in eye in forehead in in in HB ΣEYE ΣEYE ΣPARANASAL ΣUPPERLIP ΣMOUTH VMGs VMGi closure lifting nose wrinkling lip puckering smiling 1 -31.86 ± 8.54 12.73 ± 4.83 -28.08 ± 9.49 -22.89 ± 8.29 63.48 ± 21.27 34.10 ± 6.89 103.02 ± 9.34 2 -23.23 ± 7.80 11.11 ± 6.90 -17.94 ± 6.69 -17.60 ± 1.50 42.64 ± 17.82 24.37 ± 5.12 82.94 ± 11.47 3 -19.84 ± 4.76 8.51 ± 8.72 -12.84 ± 6.07 -16.49 ± 8.73 29.41 ± 23.81 19.98 ± 4.08 67.41 ± 14.72 4 -11.77 ± 6.20 9.27 ± 2.95 -9.95 ± 4.83 -10.71 ± 3.87 26.32 ± 5.14 14.07 ± 2.41 43.52 ± 11.45 5 -5.70 ± 3.18 7.69 ± 3.96 -3.88 ± 9.72 -7.12 ± 3.45 21.55 ± 10.92 9.64 ± 1.21 31.73 ± 5.33 6 -0.77 ± 4.53 1.33 ± 2.26 -3.86 ± 5.24 -4.01 ± 8.52 5.58 ± 6.07 4.94 ± 2.38 18.03 ± 7.10 r 0.792 -0.485 0.705 0.614 -0.685 -0.846 -0.937 p <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

Table XXI: Percent change of the best measures for each facial movement within HB grade Mean ± SD. r = Pearson correlation coefficient; p = p value. From Dulguerov et al. [145], without permission.

Side Repetition HB Measure Movement p % p % p % ΣEYE Eye closure 0.102 1 0.933 0.5 < 0.001 98.5 ΣEYE Forehead lifting 0.226 6 0.608 5 0.001 89 ΣPARANASAL Nose wrinkling 0.183 0.7 0.752 0.3 < 0.001 99 ΣUPPERLIP Lip puckering 0.302 1 0.388 3 < 0.001 96 ΣMOUTH Smiling 0.496 1 0.400 2 < 0.001 97 VMGs All 0.724 0.1 0.778 0.2 < 0.001 99.7 VMGi All * * 0.687 0.1 < 0.001 99.9

Table XXII: ANOVA analysis of the total variability of the best measure for each facial movement and the global facial values. p = p value. "%" represents the proportion of the total variability due to the given variable (side, repetition, and HB grade), expressed as a percent. The sum of the % across each row is 100. From Dulguerov et al. [145], without permission.

Figure 47: Box-plot of the best measures for each movement against the HB grade. Next page. X axis: HB grade. For normal subjects (HB = I), N= 5 subjects × 4 days × 3 repetitions × 2 sides = 120; for patients (HB = II to IV) N = subjects × 3 repetitions. Y axis: percent change relative to rest. A) ΣEYE change during eye closure, B) ΣEYE change during forehead lifting, C) ΣPARANASAL change during nose wrinkling, D) ΣUPPERLIP change during lip puckering, E) ΣMOUTH change during smiling. From Dulguerov et al. [145], without permission.

162

A B

C D

E

163

Figure 48: Box-plot of the HB grade against VMGs (top) and VMGi (bottom). For normal subjects (HB = I) in the VMGs plot, N=5 subjects × 4 days × 3 repetitions × 2 sides=120 and in the VMGi plot, N=5 subjects × 4 days × 3 repetitions = 60. For patients in both plots (HB II to VI), N=subjects × 3 repetitions. From Dulguerov et al. [145], without permission.

164 5.2. Gustatory sweating – normative data Using the blotting paper technique, repeated measurements with the same stencil showed the precision of the measure to be 2 mg. This corresponds to 2 μl. The results of the application of known quantity of a 0.9% NaCl solution on the stencils of blotting paper and those of the ISPH method are shown in Table XXIII and Figure 49.

The amount of sweat collected with the blotting paper technique is pretty close to the amount actually applied (Table XXIII). One exception is 2 μl, an amount, close to the resolution of the method. The correlation between the amount measured with the blotting paper technique and the surface calculated with the ISPH is excellent (Pearson correlation coefficient: 0.998; p<0.001).

Sweat quantity measured Sweat area measured with Quantity of solution with the blotting paper ISPH technique AVERAGE S.D. AVERAGE S.D. 2 μl 2.38 2.13 1.24 0.22 5 μl 4.00 0.62 1.94 0.10 10 μl 9.89 0.90 4.04 0.18 25 μl 24.98 0.53 8.34 0.63 50 μl 50.06 0.56 15.75 0.60 100 μl 99.66 2.97 28.88 1.48 250 μl 244.04 2.70 60.96 2.15

Table XXIII: Normative data of gustatory sweating From Dulguerov et al. [144], without permission.

165 250 60 ISPH Blotting paper 200 50

40 150 l μ 2

cm 30 mg ~ 100

20

50 10

0 0 2 5 10 25 50 100 250 Microliters

Figure 49: Normative data for the blotting paper and the ISPH methods From Dulguerov et al. [144], without permission.

166 5.3. Parotidectomy - General data During the 4 years of the study period, 73 patients had a parotidectomy. Two patients with a complete preoperative facial nerve paralysis (HB 6) were excluded and one patient refused to participate. In all, 70 patients were enrolled.

The population was composed of 40 males and 30 females for a male-female ratio of 0.57. The average age was 50 ± 17 years, with a range of 12 to 83 years. The right side was involved in 37 cases and the left in 33.

20

18

16

14

12

10

Frequency 8

6

4

2

0 20 30 40 50 60 70 80 >80 Age

Figure 50: Histogram of age distribution

The parotidectomy operation performed was a superficial parotidectomy in 43 cases (63%), a total parotidectomy in 26 cases (36%), and a radical parotidectomy in 1 case (1%). For the data analysis, total and radical parotidectomies were grouped together. A neck dissection was performed in 6 cases (7%), a supraomohyoid selective neck dissection in 5 cases, and a radical neck dissection in 1 cases.

The average duration of the procedure was 148 ± 57 minutes.

167 The pathological diagnosis was a benign process in 62 cases and a malignancy in 8 cases. Benign processes are divided in adenomas (48 cases), non-epithelial tumors (2 cases), tumor-like lesions (9 cases), and infections (3 cases). The adenoma group is composed of 37 pleomorphic adenomas (52.9%), 9 adenolymphomas, and 2 other monomorphic adenomas (1 basal cell and 1 canalicular). The tumor-like lesions are composed of 6 cysts (3 lymphoepithelial cysts, 1 lymphoepithelial cyst associated with AIDS, and 2 salivary duct cysts) and 3 parotid adenopathies. The infection group is composed of 1 parotid abscess, 1 parotid tuberculosis, and 1 chronic sialadenitis. The malignant lesions are divided in 6 carcinomas and 2 melanomas. The detailed distribution of parotidectomy specimen diagnosis is given in Table XXIV.

The average size of the parotid lesions removed was 2.4 ± 1.2 cm. The distribution of lesion sizes is shown in Figure 51. For the data analysis, 3 groups were considered: < 3 cm (52 or 74%), 3 to 5 cm (13 or 19%), and > 5 cm (4 or 6%).

30 27 25 25

20

15 13

10

5 5

0 < 2 cm 2 to 3 cm 3 to 5 cm > 5 cm

Figure 51: Histogram of the size of parotid lesions

168

NAME Number %

70 100 % 1. ADENOMAS 48 68.6 % 1.1. Pleomorphic adenoma 37 52.9 % 1.2. Monomorphic adenoma 11 15.7 % 1.2.1. Adenolymphoma (Whartin tumor) 9 12.9 % 1.2.2. Others: - Myoepithelioma - Basal cell adenoma - Oncocytoma 2a 2.8 % - Canalicular adenoma - Sebaceous adenoma - Ductal papilloma - Cystadenoma 2. CARCINOMAS 6 8.6 % 2.1. Acinic cell 1 1.4 % 2.2. Mucoepidermoid carcinoma 2 2.9 % 2.3. Adenoid cystic carcinoma 1 1.4 % 2.4. Adenocarcinoma 1 1.4 % - Polymorphous low grade - Basal cell - Papillary cystadenocarcinoma - Mucinous - Adenocarcinoma NOS (Not Otherwise Specified)

2.5. Carcinoma in pleomorphic adenoma 1 1.4 % 2.6. Squamous cell carcinoma 0 2.7. Undifferentiated carcinoma 0 3. NON-EPITHELIAL TUMORS Angiomas, Lipomas, Neurogenic tumors, 2b 2.9 % Mesenchymal tumors, Sarcomas 4. MALIGNANT LYMPHOMAS 0 5. SECONDARY TUMORS 2c 2.9 % 6. UNCLASSIFIED TUMORS 7. TUMORS-LIKE LESIONS Sialoadenosis; Oncocytosis; Necrotizing sialometaplasia; Benign lymphoepithelial lesion; d 12.9 % Salivary gland cysts (mucoceles, salivary duct, 9 lymphoepithelial, dysgenetic); Chronic sclerosing sialadenitis; Cystic lymphoid hyperplasia in AIDS 8. INFECTIONS 3 4.3 %

Table XXIV: Histology of parotidectomy specimen (a) Other monomorphic adenomas: 1 basal cell and 1 canalicular; (b) Non-epithelial tumors: 1 lipoma and 1 neurinoma; (c) Secondary tumors: 2 metastatic melanoma; (d) Tumor-like lesions: 6 cysts, 3 parotid lympadenopathies.

169 5.4. Parotidectomy complications - facial nerve paralysis

5.4.1. Classification according to the House-Brackmann scale

5.4.1.1. Entire population Clinical evaluation using the House-Brackmann scale (HB) shows that all 70 patients had normal preoperative facial nerve function. The average score on postoperative day 7 was 1.43 ± 0.90. The facial function was normal in 51 patients (73%) and close to normal (HB 2) in 13 patients (18.6%).

Long-term function was evaluated in 67 patients, because 3 patients with abnormal postoperative facial function have not reached the minimal follow-up delay of 6 months. The facial function was normal in 64 (96%), with 2 patients having a HB score of 2 (Table XXV). One of the patients had a 3.5-cm melanoma parotid lesion and had several small peripheral branches cut. The second had a 4.5-cm abscess of the lower portion of the parotid gland and the marginal mandibular nerve was inadvertently sectioned. Despite an epineural neurorraphy the patient never regained entirely normal facial nerve function. The final patient with long-term facial nerve deficit had a high-grade adenocarcinoma and underwent a radical parotidectomy, with sacrifice of the facial nerve. Therefore, only patients with a section of a portion of the facial nerve had persistent deficits. The average score in the long-term population was 1.09 ± 0.51.

House-Brackmann Postoperative day 7 Long-term ( > 6 months) score 1 51 72.9% 64 95.5% 2 13 18.6% 2 3.0% 3 4 5.7% 0 0.0% 4 0 0.0% 0 0.0% 5 1 1.4% 1 1.5% 6 1 1.4% 0 0.0% Total 70 100.0% 67 100%

Table XXV: Facial function scores of the entire population

In view of the paucity of persisting facial nerve function deficits, the remainder of the facial nerve data will be limited to postoperative data.

170 5.4.1.2. Role of sectioning of facial nerve branches In 8 patients, a portion of the facial nerve was cut during the procedure. Only 2 of these patients had a normal postoperative HB score. The average post-operative HB score in patients with cut facial nerve branches is 2.63 vs. 1.27 for patients with intact facial nerve. The average long-term HB score in patients with cut facial nerve branches is 1.63 vs. 1.02 for patients with intact facial nerve (p=0.036).

100% 1 1 90% 11

80%

70% 3

60% HB 6 50% HB 5 HB 3 40% 49 2 HB 2 30% HB 1

20% 2 10%

0% NO YES

Figure 52: Postoperative HB scores according to whether facial nerve branches were sectioned From Dulguerov et al.[137], without permission.

171 5.4.1.3. Patient's age and postoperative facial function

The correlation between the patient’s age with the postoperative HB score did not reach statistical significance (Figure 53).

6

5

4

3 HB score

2

1

0 0 102030405060708090 Age

Figure 53: X-Y plot of patient's age vs. postoperative HB score

172 5.4.1.4. Type of parotidectomy and postoperative facial function The distribution of the postoperative HB scores according to the type of parotidectomy is shown in Figure 54. The 43 patients who underwent a superficial parotidectomy had an average score on postoperative day 7 of 1.21 ± 0.46. The facial function was normal in 35 (67.3%) patients and close to normal (HB 2) in 7 (27%). The 27 patients who underwent a total parotidectomy had an average score on postoperative day 7 of 1.77 ± 1.28. The facial function was normal in 16 (33%) patients and close to normal (HB 2) in 6 (25%). The difference in terms of HB scores between superficial and total parotidectomy is significant (p=0.035).

100% 1 4 90% 7 3 13 80%

70% 6

60% HB 6 50% HB 5 HB 3 40% 35 51 HB 2 30% 16 HB 1

20%

10%

0% Superficial Total parotidectomy Total parotidectomy

Figure 54: Bar chart of the postoperative HB scores according to the type of parotidectomy From Dulguerov et al.[137], without permission.

173 5.4.1.5. Histopathology and postoperative facial function The distribution of the postoperative HB scores according to the histopathology of the parotidectomy specimen is shown in Figure 55. The 37 patients with pleomorphic adenoma had either normal (HB1: 28 patients – 76%) or near normal (HB2: 9 patients – 24%) postoperative facial function. The 11 patients with monomorphic adenoma had either normal (HB1: 10 patients – 91%) or near normal (HB2: 1 patients – 9%) postoperative facial function. Both patients with benign non-epithelial tumors had normal (HB1) postoperative facial function. The 9 patients with tumor-like lesions had either normal (HB1: 7 patients – 78%) or near normal (HB2: 2 patients –

22%) postoperative facial function.

adenoma adenoma tumors lesions

Pleomorphic Monomorphic Non-epithelial Tumor-like Infections Cancer

0%

1

3

20%

HB 1

28

7 40%

HB 2 1

10

HB 3 2

HB 5

60%

HB 6 2

2

80% 1

9 2

1

1 100%

Figure 55: Postoperative HB scores according to the histopathology From Dulguerov et al.[137], without permission.

Patients with infections (3) had less optimal results: only 1 patient (33%) had a normal postoperative facial function, while 2 patients had a HB score of 3 (67%). One of these had a large lesion (5-cm) of tuberculosis involvement of the parotid. The second patient had also a large (4.5- cm) intra-parotid abscess, during the resection of which the inferior division of the facial nerve was

174 sectioned. Despite an epineurial neurorraphy the patient never regained normal long-term facial nerve function. This case is the only non-cancerous lesion with long-standing facial motor deficit.

Patients with primary (6) or secondary (2) parotid malignant lesions (8) also fared less well than those with benign lesions. Only 3 patients (37.5%) had normal postoperative facial function, while the remaining 5 patients had HB scores of 3 or worse. One of these patients had a facial nerve sacrifice with a sural nerve graft for an adenocarcinoma.

Grouping of lesions in benign tumors, infections, and malignant lesions is shown in Figure 56. In the 59 patients with benign tumors, the average postoperative score was 1.20, the facial function was normal in 47 (80%) patients, and close to normal (HB 2) in 12 (20%). Therefore, no important facial nerve deficit (HB > 2) was found in patients operated on for benign tumors. Patients with infections had an average postoperative score of 2.33 and those with cancer 2.75. The difference in terms of HB scores between these 3 groups is significant (p<0.001).

100% 1 90% 12

80% 1

70% 2 2 60% HB 6 HB 5 50% HB 3 1 HB 2 40% 47 HB 1

30%

20% 3 1 10%

0% Benign tumors Infections Cancer

Figure 56: Postoperative HB scores according to three histopathologic groups From Dulguerov et al.[137], without permission

175 5.4.1.6. Size of lesion and postoperative facial function The distribution of the postoperative HB scores according to the size of the parotid lesion removed is shown in Figure 57. The 52 patients with lesions smaller than 3 cm had an average postoperative score of 1.29 ± 0.49. The facial function was normal in 38 (73%) patients and close to normal (HB 2) in 13 (25%). The 13 patients with lesions between 3 and 5 cm had an average postoperative score of 1.61 ± 1.26. The facial function was normal in 10 (77%) patients , the remaining 3 having HB scores worse than 3. In lesions bigger than 5 cm, the average postoperative score was 2.4 ± 2.2. A correlation factor of 0.27 was found and this was significant (p=0.025)

100% 1 1 90% 1 13 2 80%

70% 1

60% HB 6 50% HB 5 HB 3 40% 38 10 HB 2 30% 3 HB 1

20%

10%

0% < 3 cm 3 to 5 cm > 5 cm

Figure 57: Postoperative HB scores according to the size of the lesion From Dulguerov et al.[137], without permission

176 5.4.1.7. Duration of parotidectomy and postoperative facial function The average duration of the procedure was 148 ± 57 minutes (Figure 58). Patients with normal facial function had an average duration of 137 ± 50 minutes, patients with a HB score of 2 had an average duration of 147 ± 43 minutes, while patients with HB scores > 2 had an average duration of 240 ± 76 minutes. The duration of parotidectomy was well-correlated with the postoperative HB score (r=52; p < 0.01).

6

5

4

3 HB score

2

1

0 0 50 100 150 200 250 300 350 400 Duration of surgery [min]

Figure 58: X-Y Plot of the relation between duration of surgery and postoperative HB scores From Dulguerov et al.[137], without permission

5.4.1.8. Type of intraoperative monitoring technique and postoperative facial function The distribution of the postoperative HB scores according to the size of the parotid lesion removed is shown in Figure 59. Patients monitored with the balloon mechanical transducer had an average postoperative score of 1.60 ± 1.17. The facial function was normal in 24 of 35 patients (69%) and close to normal (HB 2) in 6 patients (17%). Patients monitored with the EMG-based device (Neurosign) had an average postoperative score of 1.26 ± 0.51. The facial function was normal in 27 of 35 patients (77%) and close to normal (HB 2) in 7 patients (20%). The difference did not reach statistical significance.

177 100% 1 4 90% 3 7 80% 13 6 70%

60% HB 6 50% HB 5 HB 3 40% 27 HB 2 24 51 30% HB 1

20%

10%

0% Balloon Neurosign Total

Figure 59: Postoperative HB scores according to the type of monitor. From Dulguerov et al.[137], without permission

5.4.1.9. Patients with poor postoperative facial function Only 6 patients (8%) in this series with postoperative (day 7) had HB scores worse than 2. Their main characteristics are shown in Table XXVI. The majority of these patients had a total parotidectomy, a diagnosis of either infection or cancer, and large lesions.

Patient Age Parotidectomy Histopathology Lesion size Monitoring device

AL 68 Superficial Infection – tuberculosis 5.0 Balloon

BM 57 Radical Cancer – Adenocarcinoma 5.0 Balloon

LC 73 Total Cancer – Melanoma 1.5 Neurosign

HR 64 Total Infection – Abscess 4.5 Balloon

RW 65 Total Carcinoma ex pleomorphic adenoma 3.5 Balloon

GSL 51 Total Cancer – Adenoid cystic 4.5 Balloon

Table XXVI: Main characteristics in patients with postoperative HB score > 2 From Dulguerov et al.[137], without permission

178 5.4.2. Videomimicography results

The correlation between the global VMG values with the HB grades was excellent (Pearson σ = -0.974; p < 0.001). The data are similar to those previously described with the HB grade (§ 5.4.1.) and only presented in a summary table here (Table XXVII).

Variable Values VMGi average VMG standard deviation Statistical test p

Age 89.11% 18.26% Pearson correlation .199

M 89.49% 17.06% Sex Student t-test .841 F 88.60% 20.04%

Superficial 93.67% 9.3% Type of surgery Student t-test .007 Total 81.85% 25.62%

No 91.90% 13.92% Branches sectioned Student t-test .001 Yes 67.50% 31.57%

Benign 94.49% 7.52%

Histology Cancer 77.29% 28.57% One way ANOVA < 0.001

Infection 66.33 30.14%

Diameter of the lesion 89.11% 18.26% Pearson correlation .04

Duration of the procedure 89.11% 18.26% Pearson correlation <0.001

Balloon 85.17% 22.98% Type of monitoring Student t-test .07 Neurosign 93.05% 10.82%

Table XXVII: Statistical relations between parotidectomy variables and VMG index Only the data at postoperative day 7 (transitory facial paralysis) are presented. Also, only the global VMG value (VMGi) is tabulated.

179

5.5. Parotidectomy complications – Frey syndrome In the 70 patients studied, 46 patients had an implant and 24 patients had no implant placed. The 46 implants placed were 7 Ethisorb®, 7 lyophilized dura, and 32 e-PTFE sheets. Only 60 patients could be evaluated one year after the surgery, because 3 are dead or lost to follow-up (all 3 in patients with no implant) and 7 have a follow-up shorter than 12 months (all e-PTFE).

The clinical evaluation showed that 5 out 52 (10%) patients consulted with complaints of gustatory sweating. All of these patients had no implant placed. Therefore, the incidence of patients consulting with gustatory sweating symptoms is 24% in the non-implant group and 0% in the other groups (p<0.001). Seven patients (13%) gave a positive history on questioning: 6 in the group without Frey-protection barrier and 1 patient with a lyophilized dura barrier. Overall, 11 out 21 (53%) patients without Frey-protection barrier had a positive clinical evaluation for Frey syndrome (Figure 60). The incidence in patients with an implant was 1 in 39 or 2.6% (p<0.001).

100% 1 1 5

80%

60% 6

72538 6 40%

10 20%

0% Lyophilized dura Ethisorb e-PTFE Any barrier No barrier

No Yes - Questions Yes - Consultation

Figure 60: Distribution of the grades of the clinical Frey syndrome evaluation From Dulguerov et al. [142], without permission

180 All patients with a clinical Frey syndrome tested positive on the objective tests. Objective Frey evaluation tests were positive in 24 of 60 patients (40%). Tests were positive in 16 of 21 (76%) patients without Frey-protection barrier, and in 8 of 39 patients (20%) with an implant. This difference was highly significant (p < 0.001). In patients with an implant, objective Frey testing was positive in 5 of 7 (71%) patients with a lyophilized dura, 1 of 7 (14%) patients with Ethisorb, and 2 of 25 (8%) patients with e-PTFE sheets (Figure 61). An exact Fisher test on the whole group is highly significant (p<0.001). Considering Ethisorb® and e-PTFE data, 3 of 32 (9.4%) patients tested positive. The area of sweating was always limited (Figure 62) and located away from the parotid area, either anteriorly towards the oral commissure, superiorly in the temporal hairline, or posteriorly behind the ear. Probably the implant was not big enough or was not positioned correctly to cover the entire exposed skin.

100% 2 1 8

80%

5 60% 16

23 6 40% 31

20% 2 5

0% Lyophilized dura Ethisorb e-PTFE Any barrier No barrier

No Yes

Figure 61: Objective Frey syndrome evaluation From Dulguerov et al. [142], without permission

181

Not only was gustatory sweating more frequent without barrier or with lyophilized dura, the intensity of the sweating was also significantly more important (Figure 62).

20

18

16

14

12 12.40 10 cm2 8

6 7.27

4

2 1.08 1.59 0 Lyophilized dura Ethisorb e-PTFE Any barrier

Figure 62: Surface of Frey syndrome using the ISPH method From Dulguerov et al. [142], without permission

There was no statistical relation between, on one side, either clinical Frey syndrome or the ISPH test results, and on the other side, patient's age, type of parotidectomy, size of the lesion removed, or the histopathological diagnosis.

182 5.6. Parotidectomy – wound complications All but 3 patients had immediate postoperative anesthesia by history (96%).

Postoperative hematoma was present in 5 patients (7.1 %). Postoperative seroma was present in 4 patients (5.7 %). The incidence of hematoma and seroma are not statistically different among the different implants used (Table XXVIII and Figure 63). Postoperative salivary fistula was present in 15 patients (21.4 %), 2 of which also had a postoperative hematoma (Figure 64). Postoperative salivary fistula was most frequent after Ethisorb mesh implants (4 patients – 57%), followed by e-PTFE sheets (8 patients – 25%). The incidence of postoperative salivary fistula was statistically different between the implant groups (p=0.035). Overall, 21 patients (30%) had a postoperative wound collection of some kind.

All parotid fistulas eventually closed with conservative treatment. In two patients the e- PTFE was exposed at the wound edges and was pulled out easily, without any anesthesia and with minimal pain reported by the patients. The average duration of drainage of parotid fistula cases was 20 ± 11 days.

The relationship between parotid collection and different implants used in the prevention of Frey syndrome is shown in the Table XXVIII. Once the initial period was over, no long-term complications of any kind were encountered.

Wound complication Lyophilized dura Ethisorb ® Gore-Tex® Nothing TOTAL

Seroma 14.3 % 0.0% 6.3 % 4.2 % 5.7 %

Hematoma 0.0 % 14.3% 9.4 % 4.2 % 7.1 %

Salivary fistula 0.0 % 57.1% 25 % 12.5 % 21.4 %

Any collection 14.3 % 71.4% 31.3 % 20.8 % 28.6 %

Table XXVIII: Frequency of the different wound collection complications and type of implant Modified from Dulguerov et al. [142].

183

100% 3 90% 8 12 80% 70% 4 60% 50% 7 21 40% 24 34 30% 20% 3 10% 0% Lyophilized Ethisorb Gore-Tex Any barrier Nothing dura No Yes

Figure 63: Incidence of postparotidectomy fistula and type of implant

100% 1 90% 5 10 80% 16

70% 5 60%

50% 6 40% 19 22 30% 30

20% 2 10%

0% Lyophilized Ethisorb Gore-Tex Any barrier Nothing dura

No Yes

Figure 64: Incidence of parotid wound complications (overall) and type of implant

184 5.7. Frey syndrome treatment with botulinum toxin All patients treated with botulinum toxin had clinically symptomatic Frey syndrome and requested some form of treatment. All complained of profuse sweating that recurred with each meal. Two patients also complained of gustatory flushing.

The surface area involved before treatment was 32 ± 23 cm2 (range: 49 to 3.5), according to the ISPH. The amount of sweating before treatment in this population was 67 ± 23 mg (range: 104 to 65) with the blotting paper technique.

The amount of botulinum toxin injected varied from 0.3 to 1.5 ml, which corresponds to 15 to 75 I.U. No side effects were reported by the patients or noticed by the observers.

After injection with botulinum A toxin, the surface area was 0.45 ± 0.41 cm2 (range: 0.74 to 0.02) with the ISPH and the amount of sweating was 1.5 ± 0.7 mg (range: 2.3 to 1.1) with the blotting technique [143]. The average amount change was thus 77 ± 14 mg (Figure 65) and the average surface change was 14.7 ± 17.4 cm2 (Figure 66). Both these changes were statistically significant (p<0.05 and p<0.001).

5.8. Recurrences So far, we have had 4 recurrences in the 70 postparotidectomy patients. Two recurrences are among the 8 patients with malignancies (25%). In both these patients the recurrence was treated with an extensive resection, including subtotal petrosectomy, total TMJ resection and reconstruction, and an infratemporal fossa resection [442]. One these patients died accidentally (mountain climbing), two years after this salvage surgery, without having had any apparent recurrence. The second patient moved overseas and was lost to follow-up. Among the remaining 6 patients with malignancies, 4 are alive and well, 1 with parotid melanoma is dead of disease (local, regional, and distant diseases), and 1 patient has moved away and is lost to follow-up.

Two recurrences are in the benign lesion group, for an incidence of 3.2%. Both patients had a pleomorphic adenoma. The size of the original tumor was 2.8 cm and 4.0 cm, respectively. One patient had a superficial and one patient had a total parotidectomy. One patient had a single recurrent nodule excision under local anesthesia as treatment of his recurrence. The second patient has multiple nodules along the scar line. She underwent 2 procedure: the first was an attempt to

185 resect the nodules and dissect the facial nerve. Since the facial trunk was found embedded in the recurrence, nerve sacrifice and grafting was performed. Her facial nerve grading is 4.

60 BEFORE

50 AFTER

40

30 cm2

20

10

0 ISPH

Figure 65: Frey syndrome quantity before and after treatment with botulinum toxin

0.100

0.090

0.080

0.070

0.060

0.050 g ~ ml 0.040

0.030

0.020

0.010

0.000 Blotting Paper

Figure 66: Frey syndrome surface before and after treatment with botulinum toxin

186 187 6. DISCUSSION Since surgeons like any performers have little objectivity regarding their own results, we felt compelled to use objective methods for the evaluation of parotidectomy complications. While the detection of a preoperative facial nerve asymmetry has obvious implications, the necessity of such objective tests outside a study protocol remains debatable. Nevertheless, a photographic record of either facial function or facial sweating, analogous to cosmetic surgery records, has several non-legal aspects. The examination of postoperative facial movements is a great reward when all is well and quite instructive if there is a deficit. The facial deficit is usually limited to a certain facial branch or territory; this should direct the surgeon to reexamine the operation records in search of a traumatic factor responsible for this deficit. Was a branch sacrificed? Was the dissection especially difficult in this area? Such a approach should help the surgeon improve his technique, his understanding of the physiopathology of postoperative facial paralysis, and his overall future results.

After a review of available methods for facial motor function analysis (§ 2.1.2 to 2.1.4) and for sweating measurement (§ 2.2.5), it became obvious that we needed do develop new methodology.

6.1. Videomimicography To develop any test, the first prerequisite is to decide exactly what should the test measure. For objective facial nerve evaluation methods (OFNEM) this means to specify what facial movements should be studied, how should the movements be performed, and what should be measured for each movement. Only than can issues about the way to perform these measures be raised [145].

6.1.1. What facial movements?

Any published facial evaluation method, being either subjective or objective, has assumed that the production of facial movements is a reliable representation of facial neuromuscular function. We also assumed that a variation of the standard facial movements used in "clinical" evaluation is what should be tested. While the movements we choose cover most of the facial mimetic musculature, from the forehead to the chin, the pertinence of these movements should be addressed by a correlation with some form of facial disability evaluation, of which disability questionnaires [236; 327; 380; 531; 547] are one example. Such a study is yet to be performed.

188 6.1.2. How should the movements be performed ?

Once the facial movements to be assessed are determined, it remains to specify how should these movements be performed. Only few OFNEM have specified that maximal contractions were specifically requested for each facial movement [258; 369]. Experimental arguments favoring the use of maximal contraction can be derived from Burres' studies [72]: the measures assessed for soft eye closure had a higher variability than those for tight eye closure. Probably because it seems the only simple way to standardize the production of facial movements, maximal contraction should be recommended for any future OFNEM. During videomimicography (VMG), we not only requested the maximal possible effort, but also actively verbally stimulated the subjects during the test.

6.1.3. What should be measured ?

In previous reports, few studies have compared different measures to assess the best measure for each movement. In his pioneering study Burres [72] recommended the following distances for different facial movements: forehead movements – FIo, eye closure – FIo and NaIo, nose wrinkle – AM, NaA, kissing M to lateral canthus, smiling M to mid mouth (Table VI). Frey et al. [186] using a sophisticated setup with 4 different camera and somewhat different movements and landmarks reached similar conclusions.

Previous linear measurement OFNEM have all used facial distances as measures, while image subtraction methods [356; 377], although based on a different technology, could be regarded as measuring areas. In our study, we studied both distance and area measures, at the same time. Ten distances and five areas were analyzed in each frame (18 frames per subject). The best measure for each facial movement studied was determined: ΣEYE for eye closure and forehead lifting, ΣPARANASAL for nose wrinkling, ΣUPPERLIP for lip puckering, and ΣMOUTH for smiling. These measures exhibited the largest changes relative to rest and the smallest coefficient of variation. Area measures were found to better estimate the five routine facial movements than distance measures in normal subjects.

In all 5 movements, the measures with the highest changes were located close to the moving facial area, while those located far from the moving facial area had low changes or no real change, a conclusion that could be also drawn from Burres [72] and Frey [186].

189 Another issue is whether movement-specific data should be provided or an overall score is of interest. In general, probably for ease of communication and the necessary correlation with facial grading scales, a general score is sought [73; 171; 250; 356; 369; 416]. The House-Brackmann grading system (HB) [243], because of its endorsement by the Facial Nerve Disorders Committee of the American Academy of Otolaryngology – Head and Neck Surgery, has enjoyed a large popularity and has become a standard system for the evaluation of facial nerve paralysis. However, it is subjective, semi-quantitative and is difficult to use when only one branch of the facial nerve is injured (for example, after facial surgery), because the grade does not provide information about specific facial regions [137].

In this study, a VMG score (VMGs) and a VMG index (VMGi) were created for overall evaluation of facial function, by summing the best measure for each facial movement and adding, somewhat arbitrarily, a coefficient according to the importance of each movement in facial expression. VMGs is a summary of the best measures of five routine facial movements, in which 40% is accounted by eye closure, 20% by smiling, 20% by lip puckering, 10% by forehead lifting, and 10% by nose wrinkling, because facial asymmetry is most obvious and functionally disfiguring in the eye and mouth area [236; 327; 380; 531; 547]. VMGi is calculated by comparing the VMGs of the paralyzed hemiface with that of the healthy hemiface. The best measures for each facial movement VMGs, and especially VMGi (r=-0.937) had a linear correlation with the HB grade. Theoretically, VMGi decreases from 100 in normal subjects to 0 in complete paralysis. For normal subjects, VMGi had a mean value of 103.02 and for patients with total paralysis, the average VMG index was 18.03. This may be explained by the slight movement of paralyzed hemiface drawn by the healthy one, although two points (Li and Ls) have been adjusted in smiling movements.

6.1.4. How should the facial measurements performed?

The next question to address is how should the measurements be done. An ideal OFNEM, in order of importance, should: 1) not impede facial movements, therefore the face should not be touched during the movements or for the measurements; 2) be reproducible for a given individual, both in normal and pathologic cases; 3) provide synchronous data from the left and right side of the face, for comparison; 4) not require the observer to make the measurements, avoiding manipulation errors and bias; 5) be rapid, simple and low cost; 6) well tolerated by the patients; 7) provide absolute values, not just percentages; 8) be stored in some form for later comparison, evaluation by other examiners, or further studies; and 9) not require markings on the face [140].

190 6.1.4.1. Not impede facial movement and not touching the face. While image subtraction methods have the inherent advantage of not requiring a contact with the patient's face, for linear measurements, most authors have so far used a manual technique [72; 171; 186; 416], which requires touching the patient’s face. Several authors have used still photographs [258], tape images [369], or a complicated digital setup [186], but a simple digital technique has yet to be successfully applied.

Videomimicography (VMG) is a measurement system, in which the patient face is not touched. A digital video is obtained and frames or movies can be either directly visualized or fed in a computer for analysis.

6.1.4.2. Reproducibility for a given individual, both in normal and pathologic cases. Few previous studies have assessed reproducibility in OFNEM. Wood et al. [555], using a microscaling technique, examined in 11 normal subjects performing two facial movements (brow lift and smiling), the average variability: test-retest was 4 and 5%, day-to-day was 5 and 6%, side-to- side was 6 and 14%, and intersubject variability was 25 and 23%. Neely et al. [378] used a general linear model to examine the variability in their image subtraction OFNEM. The model predicted between 82 and 95% of the observed variability and 70 to 85% of the variability was due to intersubject differences, while intrasubject variability was less than 2%. This study used only one repetition the same day and did not explore day-to-day repeatability.

Our analysis of variance is in fact a general linear model in which we examined the contribution to the total variability of same day test-retest (3 repetitions), day-to-day retest (4 days), side-to-side (2 hemifaces), and intersubject (5 normal subjects) variability. Our findings confirm the results of Wood et al. [555] and Neely et al. [378] that intrasubject variability is low compared to intersubject differences. In general, intersubject differences are responsible for 80-95% of the total variation. Therefore, it could be concluded that intrasubject variability of VMG, as well as other OFNEM is very good. We tend to disagree with Wood et al. [555] that an intersubject variability of about 25% prevents the development of OFNEM as valid tests.

While the reproducibility in patients with facial paralysis deserves a study in itself, the different OFNEM proposed have yet to address this point. Usually it is assumed that if the test is reproducible in normal subjects it should also be reproducible in pathologic cases.

We have assessed the variability in terms of the side of the face, the same day test-retest and the degree of facial paralysis as scored by the HB scale (Table XXII). Retest on different days is

191 questionable on theoretical grounds because of the natural evolution of patients with facial paralysis. Here the HB grade was responsible for 90 to 99% of the total variability. Therefore, it could be concluded that VMG is well suited to assess facial paralysis, with limited variability from other factors.

6.1.4.3. Synchronous data from the left and right side of the face. In VMG and in other OFNEM based on videorecordings of the facial movements, both sides of the face could be evaluated simultaneously. OFNEM based on direct measures on the patients face usually use asynchronous facial movements for their measurements.

6.1.4.4. Automatic measurements to avoid manipulation errors and observer bias. Several techniques proposed as OFNEM involve complicated manipulations by the examinee and it is questionable if they should qualify as objective [258; 555; 563]. All techniques involving direct measurements on the face [72; 171; 416] carry inherent observer bias. In other described linear OFNEM it is simply unclear how are the measurements performed [369] and sometimes what is measured [155]. Image subtraction OFNEM are by definition digital, however, the facial areas to be analyzed are rarely clearly defined [356; 378; 379] and the threshold for "color change" has remained arbitrary [356; 377; 378; 379]. Finally, some studies [250; 563], including ours, while giving the impression of being completely automatic, require at least some observer input.

Even in its present state VMG requires minimal observer intervention. A manual pointing with the mouse on the computer screen is necessary in order to obtain the coordinates of the different landmarks. With the custom-modified Osiris software, the coordinates and all required calculation are than send to a spreadsheet or another mathematical software. We are in the process of implementing an entirely digital system, however the entire VMG session has to be fed into a computer and software able to track the landmarks across frames has to be implemented.

6.1.4.5. Rapid, simple and low cost. In general, the more sophisticated and objective an OFNEM is, the more complicated it is in terms of time and cost of the involved equipment. Simple systems involving direct linear measurements on the face [72; 171; 186; 416] require little equipment, but have numerous shortcomings as previously discussed. Other linear OFNEM tend to involve either numerous time-consuming observer manipulations [258; 369] or extremely expensive equipment [186]. Image subtraction methods [356; 377] could be criticized because they require special lighting equipment

192 and ambient luminosity control, fixed subject-camera distance, long duration of the procedure (10 minutes), almost absolute head immobilization, and rather expensive computer setup.

For VMG, videotape frames have been individually fed into the computer directly from a digital video recorder. With the certain fall in prices of digital video recorders and the advent of IEEE 1394 (Fire wire) I/O ports as a computer standard the system remains relatively simple. What is required is a chair with a headrest, an eyeliner pen and remover, a digital camera and a computer. The analysis software can be freely downloaded.

6.1.4.6. Well tolerated by patients. Patient's acceptance is a major drawback for image subtraction OFNEM, where a head immobilization in a special head holder for 10 minutes [356] is required. In VMG the recording session take about 30 seconds and has been well tolerated by patients who often have spontaneously requested an evaluation to "assess their improvement".

6.1.4.7. Absolute values, not just percentages. Image subtraction methods have been expressed in number of pixels and absolute values have not been available. Simple systems involving direct linear measurements on the face [72; 171; 186; 416] can obviously obtain real distance directly, although the published results are often in percents. In VMG absolute as well as relative measures are available. However, and contrary to most of previous reports, in order to diminish errors derived from the placement of landmarks and the size of faces, relative changes of the measures were derived by comparing movement frames and rest frames.

6.1.4.8. Stored for later comparison and evaluations. This point is obvious and, with present video facilities, could only be neglected for simplicity reasons.

6.1.4.9. Not require markings on the face. Ideally, the patient face should not be touched at all. Nevertheless, all OFNEM except image subtraction techniques have used some kind of facial landmarks. Eyeliner landmarks are easy to remove with make-up removal solutions and have not been a problem.

In summary, VMG gives an objective, quantitative, and reproducible data for the evaluation of facial nerve motor function. The method is relatively simple and correlates well with established subjective standards, such as the House-Brackmann grading scale. While it will probably never

193 replace subjective methods, because of their simplicity, further developments could render this method entirely automatic.

6.2. Objective evaluation of Frey syndrome The techniques developed for the evaluation of sweating are remarkable by their simplicity. The historic origin of either technique is difficult to trace. We gave up efforts to trace previous uses of a blotting paper for quantitative evaluation of produced fluid. The ISPH method could be traced to an article by Dole and Thaysen in 1953 [126]. They used small disks in attempts to estimate the density of skin eccrine sweat glands. Curiously, little has been published recently concerning this technique, and it is unclear that it is used with any frequency. Sato, a world established expert of sweating disorders [458; 459; 460; 461], describes in 1988 [462] another modification of the Minor test [360]. This modified test still shares the numerous disadvantages of the Minor test and could be easily replaced by the ISPH described here [144].

In the calibration process with normal saline, the blotting paper technique recorded weight changes close to the amount of saline actually applied (Table XIX). With increasing amounts of saline applied, the blotting paper technique measured higher weights, and the surface of the ISPH method was larger. More important, an excellent correlation was obtained between the results of both methods (Figure 49).

The tests described satisfy most of the requirements of an ideal sweat test: quantitative test, simplicity, sensitivity, reliability, adequate dynamic range, absence of toxicity and allergenicity, easy removal from the skin of the applied agents, and low cost (see § 2.2.5). In addition, these two tests give complementary data. While the quantitative data obtained with the blotting paper method is excellent, the topographic information is sub-optimal. However, the topographic data of ISPH technique is excellent, providing a mirror image on the facial sweating area. Topographic information is essential for Frey syndrome treatments requiring localized application of the medication, such as botulinum toxin A injections (see § 5.7 and § 6.4.) [128; 143; 425].

Probably the only major disadvantage of the ISPH method in testing Frey syndrome is the lack of dynamic results. While dynamic measures might be interesting in investigation studies, their use in clinical practice and more specifically in gustatory sweating appears of limited interest. Even the proponents of the only dynamic method proposed for the evaluation of Frey syndrome, used

194 the Minor test when evaluating their patients [292; 293]. The excellent correlation of the two methods proposed can be used as an internal validation, if one wishes to use only one of these methods. The blotting paper has the advantage of being very simple to use and of actually measuring the amount of sweat produced; however, no topographic data are obtained. The ISPH method is slightly more complicated, but, in our opinion, still much simpler than the traditional Minor test. Depending on the goals pursued, the digitalization aspect can be left out, while still obtaining excellent topographic data.

We cannot compare our methods or results with other publications, since the majority of previous publications on Frey syndrome were based on subjective clinical evaluation. The few studies that used an evaluation test employed the Minor test, which does not give quantitative data. The only previous study to use a quantitative test is by Laccourreye et al. [289], which was mostly concerned with the description of a new technique. That paper, unfortunately, did not provide data in ml of sweat secreted and is therefore difficult to compare with our results.

6.3. Post-parotidectomy facial nerve paralysis Contrary to previous studies, the data presented were prospectively collected, and there was no specific patient’s selection. In addition, facial motor function was evaluated according to an established grading system [243]. Furthermore, the evaluation of the facial data was performed from videotape review and in a blinded fashion. These features make meaningful comparison of our data with previous publications difficult.

The overall, normal facial function (HB 1) was present in 73% of our patients on post- operative day 7 and in 96% on prolonged follow-up. Stated otherwise, 27% of patients had some form of postoperative facial deficit and 4% (including 1 patient with nerve sacrifice) had a long- term deficit. In recent publications, the incidence of temporary deficits was 18% for O’Brien et al. [394] and Watanabe et al. [538], 37% for Bron et al. [61], 46% for Mehle et al. [355], 52% for Ruaux et al. [450], 62% for Terell et al. [511], 65% for Laccoureye et al. [291], and 68% for Wolf et al. [554]. In the same publications, the range of long-term deficits is from 0% [538; 554] to 19% [394]. While these incidences might appear non-flattering, such high numbers probably represent the real incidence of early postparotidectomy facial deficit when sought by honest, critical and analytical observers.

195 Whether, the use of a routine continuous intraoperative EMG-based facial nerve monitoring has a significant impact on postoperative facial function will remain debatable. Two retrospective, non-randomized studies have compared EMG-monitoring vs. traditional parotidectomy. Wolf et al. [554] found HB scores > 1 in 69% of monitored vs. 75% of unmonitored patients (no statistical analysis provided). Terrell et al. [511] reported on abnormal facial function (author’s scale of deficit grading) in 44% of monitored and 62% of unmonitored patients (p=0.04). While these data are in favor of the routine use of EMG-based facial nerve monitoring, only a prospective, randomized study, after stratification for the risk factors discussed below, can settle the role of routine facial nerve monitoring in parotid surgery. Our personal opinion is that such a study is nowadays unethical. When we experienced some technical problems with the EMG apparatus [218], we rescheduled the patients until its repair.

The factors associated with a higher incidence of temporary facial nerve deficit include the extent of surgery (superficial vs. total parotidectomy), the sectioning of facial nerve branches during surgery, the histopathology, the size of the lesion, and the duration of the operation. As discussed earlier, several of these factors have been previously reported, although their exact significance remains controversial. Interestingly, two multivariate studies [291; 511] found patient’s advance age as the most important factor associated with postparotidectomy facial deficits. We were unable to confirm these findings and did not perform a multivariate analysis because of the relatively small size of our population.

While the role of these factors deserves further study, most of them cannot be directly controlled by the surgeon. More interestingly, the exact physiopathological mechanisms of postsurgical nerve paralysis are still poorly understood. Patey logically classified postparotidectomy facial paralysis as secondary to: 1) deliberate sacrifice of the facial nerve or its branches, 2) inadvertent but recognized section of the facial nerve or branches, and 3) unclear causes while the anatomical integrity of the nerve is intact [409]. In one of the rare experimental studies addressing this issue, Patey and Moffat [411] found the rabbit facial nerve “quite resistant to direct mechanical trauma,” sensitive to cooling (although the temperature was not measured), and speculated against nerve edema and for a the possible role of nerve ischemia. The role of nerve ischemia as a direct etiology can probably be safely ruled out because numerous animal experimental studies have shown peripheral nerves to be quite ischemia-resistant. Cooling is also an unlikely etiology, although data on facial nerve temperature during parotidectomy have never been published, since nerve cooling to 20 °C has repeatedly been shown to be unharmful [463].

196 Mechanical trauma can be separated in compression, crushing, and stretching (see § 2.1.5.). Apparently, peripheral nerves can withstand compressions around 100 mmHg (13 kPa, 2 lbf/in2) before the nerve microcirculation becomes impaired, resulting in a metabolic conduction block [324]. At higher and sustained pressures, focal demyelination takes place, which requires 6-12 weeks for a complete recovery [395]. Again, it seems rather unlikely that during careful parotidectomy the facial nerve will be compressed with such high pressures. Crushing of peripheral nerves with surgical forceps reliably produces a mechanical deformation of the myelin sheets, resulting in a segmental demyelination [324], which also requires 6-12 weeks for recovery. While facial nerve branches crushing could occur, it seems to be a rare phenomenon in careful parotidectomy. The most probable mechanical factor involved is nerve stretching. Peripheral nerves have been found to follow a peculiar stress-strain curve with zones of straightening, elastic elongation, followed by mechanical rupture (see Grewal et al. [208] for a recent review). Earlier data showed that rupture occurs at 38% elongation; however, more recent studies have demonstrated perineurium tears with disturbances of the intrafascicular homeostasis at elongations of 6% [208]. The resulting edema further impedes the microcirculation of the nerve and results in an unrecoverable loss of the compound action potential. During such trauma, the nerve remains grossly normal. It is easy to imagine how such nerve stretching could happen during parotidectomy.

Other possible etiologies of nerve damage include heat damage from electrocoagulators (unlikely without massive nerve twitching), damage from overzealous nerve stimulation (unlikely in view of experimental and clinical data with various functional electric stimulating implants), and damage from neurotoxic substances placed in the surgical wound (unlikely). Therefore, experimental animal data point to nerve elongation as the most probable factor involved in anatomically intact facial nerves associated with postparotidectomy facial paralysis.

The exact role of these data drawn from animal experiments not related to the facial nerve will remain speculative until parotidectomy animal experiments are carried out and their implications for human parotidectomy fully evaluated.

197 6.4. Frey syndrome: prevention, detection, and treatment This study confirms previous reports [7; 201; 278; 284; 319] on the incidence of Frey syndrome without preventive measures (see § 2.2.6.). Our results in the group without barrier showed a 53% of clinical Frey syndrome and a 75% of Frey syndrome by using an objective testing method. Our clinical Frey data seem to be slightly higher than the average of Table XV. A possible reason for this is our aggressive questioning: for the "YES-QUESTIONS" category, patients were directly and thoroughly asked about possible secretion/draining/sweating during eating. Nevertheless, it is reasonable to conclude that the incidence of clinical Frey syndrome, i.e., patients that are aware of a gustatory facial sweating, is around 40-50%.

Our objective test data in patients without preventive barrier (75%) are somewhat lower than the average incidence from the literature (87%). One possible explanation is the use of different tests. All previous investigators have used the Minor test. Although we have classified this test as objective, it is unclear 1) how technical variations (differences in the solutions concentrations, differences in the exposure time, etc.) influence the results and 2) what threshold is used for a positive test (few sweat gland drops, slight color changes, etc.). In addition, while it is possible to take pictures of Minor test hemifaces and perform the analysis on these pictures, none of the previous studies reported so. Finally, the Minor test is interpreted without any reference, since both patients’ hemifaces are usually not tested. The objective tests developed for this study, have been calibrated with known quantities of saline (see § 5.2.) [144] and have several advantages over the Minor test (see § 6.2.). The iodine-sublimated paper histogram (ISPH) does not require painting of various solutions on the face, avoiding not only patient's discomfort but also potential allergies. Because the test is well tolerated, bilateral testing is used allowing for a normal reference. While we have noticed clear differences in results when the same patients were tested on warm days and after an effort (data not shown), the differences are bilateral and therefore cancelled out in the result calculations. Finally, contrary to the Minor test, an objective analysis of data is carried out by the computer image histogram analysis, allowing to set an objective threshold for test positivity. This test is simple to perform and the technical details have been provided earlier (see § 5.2.).

The comparison of the different Frey syndrome prevention barriers used shows that they were effective in preventing clinical Frey syndrome (1 in 39 patients – 2.6%). This case occurred after placement of lyophilized dura and was located very anteriorly, close to the corner of the mouth, and probably in front of the implant placed. Then ISPH test data are examined,

198 Ethisorb®, and e-PTFE were more efficacious than lyophilized dura. While its seems logical that non-resorbable material such as e-PTFE gives better and more permanent results, the difference between 2 resorbable materials such as lyophilized dura and Ethisorb® is purely speculative. Probably some difference results from the resorption characteristics of the material, which are not known in great detail and might vary between patients. Our final impression, partially supported by the data is that the less resorbed an implant, the better barrier it is.

The resorption characteristics of the implant are to be put in perspective with the wound complications it produces. A seroma is the mildest form, since by our definition is resolves without any treatment. We had the impression that hematoma was not directly related to the implant, but might result more from operative events, such as inadequate hemostasis under hypotensive anesthesia with rebound bleeding and uncontrolled postoperative hypertensive episodes. Therefore, the most annoying wound complication is salivary fistula, which was clearly more frequent with Ethisorb® mesh sheets, but also with e-PTFE. Once the acute period (4-6 weeks) has passed, no specific problems with the implants were encountered.

The ideal Frey prevention barrier has either to remain in place permanently or to be replaced by a body fibrosis, which is dense enough to prevent the growth of parasympathetic parotid fibers towards the facial skin eccrine glands. In many respects, e-PTFE implants represent the ideal solution since they are not resorbed, and exhibit good biocompatibility and low tissue reactivity [326; 376]. However, probably because of this low and slow biointegration, e-PTFE, like the Ethisorb® seem to act as a foreign body in the postparotidectomy wound and stimulate saliva secretion. If our initial enthusiasm with e-PTFE has been somewhat chilled by these wound side effects, it remains our implant of choice until more suitable materials become available.

A review of the literature of other techniques used in preventing Frey syndrome is compiled in Table XVII. Probably the only technique of potential interest is the so-called SMAS flap technique. The SMAS flap technique has the obvious advantage of not using exogenous material. Disadvantages of the SMAS flap technique disadvantages include 1) the difficulty of using it consistently in all cases because the size of the tumor may require a sacrifice of the SMAS layer, 2) often area of SMAS are resected or transections are present in the fascia during the creation of the initial flap, 3) difficulty in using it in revision parotidectomies, and 4) greater technical difficulty during surgery. While evaluation in early studies using the SMAS flap technique was done mainly by history, recent data using the Minor test are less favorable (see Table XVII). Only a randomized trial, using an evaluation with an objective test, comparing the SMAS flap technique and the barrier

199 method advocated here can demonstrate the best method of preventing Frey syndrome during parotidectomy.

In cases of established Frey syndrome that are symptomatic, the available treatments have either been of temporary benefit or are associated with potential side effects that are more important than the symptoms to be treated. The treatment of Frey syndrome with botulinum toxin injection, following a suggestion of Drobik and Laskawi [128], appears an attractive option.

The action of botulinum toxins is that of highly specific endopeptidases that cleave three different proteins (synaptosome-associated protein or SNAP-25, synaptobrevin, and syntaxin) involved in the fusion of exocytosis vesicles with the cell membrane [363]. The action of botulinum toxin type A is on SNAP-25 [49]. While this action could occur in any cell, the extremely high specificity for cholinergic synapses is due to specific membrane proteins found only in the presynaptic endings of cholinergic synapses [98]. After binding, the toxin is internalized by endocytosis and liberated in the presynaptic cytosol. By blocking the exocytosis mechanism of the presynaptic terminal, the release of acetylcholine is inhibited. The synapse remains intact but is non-functional and for neuromuscular junctions results in muscle paralysis [52]. The recovery of neuromuscular function is due to collateral sprouting from the same or other axons and the formation of new synapses [52].

Numerous advantages can be cited for the use of botulinum toxin for the treatment of Frey syndrome. The substance is relatively well known and has been used for a variety of pathologies in the last 20 years with few side effects [255]. Its use is rather simple in the office setting, and avoids complicated procedures in the operating room. The only disadvantages are the pain associated with the needle sticking and the secondary reluctance of some patients to be stuck, especially on the face. The pain associated with the needle injection was evaluated by our patients to be minimal (2/10 on an analog visual pain scale), a result found also in other studies [46; 294]. Whether the use of local anesthetic cream (EMLA®, Astra, Sweden) will increase patient's acceptance of this mode of treatment [46] remains to be seen.

We did not experience any facial paralysis side effects of the botulinum toxin injection, as reported by others [46; 294]. It is difficult to understand how a strictly intradermal injection can result in paralysis of facial nerve branches. In addition, the membrane receptors responsible for the cholinergic specificity of botulinum toxins are located only in the presynaptic terminals and not in nerve fiber trunks. Therefore, the most plausible explanation of these rare and partial temporary paresis is an injection deep to the dermis and diffusion to facial motor end plates.

200 Before a widespread acceptance of this mode of treatment, the technical aspects in terms of botulinum toxin dilution, inter-injection distances, doses per injection site, and maximal doses per patient should be determined (Table XXIX). There seems to be an agreement on the 10-mm inter- injection distance and on an injection dose of 0.1 ml. The 10-mm inter-injection distance is supported by the data of Shaari and Sanders for neuromuscular junctions [474]: injections 10 mm away from the motor end plate band of rat tibialis anterior muscle are ineffective in inducing paralysis. In the same experimental system, increasing the dose and volume of botulinum toxin resulted in increased paralysis. A 20-fold increase in dose doubled the paralysis; however, little gain was achieved with injection doses above 5 units [474]. As far as volume is concerned, a 100-fold increase in volume was necessary to double the paralysis [474]. Studies on human sweat glands partially support these data [76].

STUDY Year Patients Botox® Inter-injection Dose per Dose per Maximal dose dilution distance injection patient (U.) used

Bjerkhoel [46] 1997 14 25 U. / ml 1 cm 0.1 ml 38 ± 12 62.5

Naumann [373] 1997 45 20 U / ml 1.5 cm 0.05 to 0.1 ml 21 ± 14 72

Laccourreye 1998 12 25 U. / ml 1 cm 0.1 ml 65 88 [294]

Laskawi [302] 1998 19 25 U. / ml 2 cm 0.1 ml 31 100

Dulguerov 1998 16 50 U. / ml 1 cm 0.1 ml 41 ± 22 75

Table XXIX: Characteristics of the botulinum toxin type A dilution and doses in publications on Frey syndrome treatment with botulinum toxin

Small volumes of concentrated botulinum toxin have the advantage of minimizing the diffusion and possible side effects. However, if side effects do occur they would be more disabling because of the higher concentration [474]. We used a concentration of 50 U per ml, with the idea that a smaller volume of injection will produce less discomfort and be more efficacious. This is probably necessary in zones with intense facial gustatory sweating. Others [46; 294] have obtained good results with a lower dilution (25 U per ml), however, and this dosage might be advantageous in patients who have a large surface of gustatory sweating. Apparently, the duration of the effect of

201 botulinum toxin on gustatory sweating is rather long lasting, with an average of 15 months [292; 302]. An actuarial estimate was recently published [292], with clinical gustatory sweating recurrence at 1, 2, and 3 years of 27%, 63%, and 92% respectively. The severity of the recurrent Frey syndrome was found to be reduced when compared to that of the initial episode and re-treatment with botulinum toxin was still effective [292].

We have the impression that the new methods that were developed for the evaluation gustatory sweating (and especially the ISPH) [144] are invaluable for the decisions of the zones to treat and the amount of botulinum toxin that should be injected. While the Minor test provides the indispensable topographic information, it is cumbersome to use [144], poorly tolerated by patients [39], does not provide a permanent record, per se, or quantitative data [144]. Also in areas of intense gustatory sweating the bluish discoloration tends to drip down the face, resulting in a smear and rendering the collection of precise topographic information difficult. With a more widespread use of the ISPH technique, the technical details of the botulinum toxin treatment for Frey syndrome can possibly be further advanced.

6.5. Wound complications The resorption characteristics of the implant are to be put in perspective with the wound complications it produces. A seroma is the mildest form, since by our definition it resolves without any treatment. We had the impression that hematoma was not directly related to the implant, but might result more from intraoperative events, such as inadequate hemostasis under hypotensive anesthesia with rebound bleeding, and uncontrolled postoperative hypertensive episodes. Therefore, the most annoying wound complication is salivary fistula, which was clearly more frequent with Ethisorb® mesh sheets, but also with e-PTFE. Once the acute period (4-6 weeks) had passed, no specific problems with the implants were encountered.

6.6. Recurrence Little can be said about recurrences from our data. We had four recurrences: two pleomorphic adenomas and two malignant tumors. The incidence for pleomorphic adenomas and cancers is respectively 3.2% and 33% (2 out of 6).

202 This 3.2% incidence of pleomorphic adenoma recurrence is within the range found in the literature, i.e., 6.7% for superficial and 8.2% for total parotidectomy (see Figure 33). However, we still feel that this number is too high for a benign tumor, especially regarding our short follow-up time (31 ± 16 months). These data will be reviewed subsequently when a longer follow-up (>5 years) is available.

It is even more difficult to analyze the factors associated with recurrence discussed earlier (see § 2.4.2.). Sufficient to say that one of the patients with a 4.0-cm pleomorphic adenoma had a tumor spill during initial surgery and she presented a classical picture of multicentric recurrences along the scar line.

203 7. CONCLUSIONS 1) Videomimicography (VMG) is a new objective, quantitative, reproducible, and relatively simple method for evaluating facial nerve function.

2) Area measures are better than distance measures in evaluating facial movements in normal

and pathological subjects. The best measures for eye closure and forehead lifting was ΣEYE,

for nose wrinkling was ΣPARANASAL, for lip puckering was ΣUPPERLIP, and for smiling was

ΣMOUTH.

3) In VMG, most of the variability in normal subjects is due to intersubject variability, while the retest variability is low.

4) All the VMG measures of facial movements and the global indexes have a good correlation with the HB grade in patients with facial paralysis.

5) The routine use of an EMG-based facial monitoring was found extremely helpful during parotid surgery.

6) In this study of unselected patients, the overall incidence of facial paralysis in this study is 27% for temporary deficits and 4% for long-term deficits.

7) Important temporary facial nerve deficits (HB>2) were not found in patients undergoing parotidectomy for benign tumors.

8) Permanent deficits were present only in patients who had a section of nerve branches.

9) Factors associated with an increased incidence of temporary facial paralysis include the extent of parotidectomy, the intraoperative sectioning of facial nerve branches, the histopathology and the size of the lesion, and the duration of the operation.

10) A review of the physiopathological factors, possibly responsible for facial nerve deficits, points to nerve stretching as the most probable etiology.

11) The iodine-sublimated paper histogram (ISPH) test for facial gustatory sweating is an accurate, reliable, easy to perform, and well-tolerated objective test. The topographic information is essential for Frey syndrome treatment involving local application of a medication. In addition, the quantitative data provided are indispensable in evaluating the results of a given treatment.

12) The incidence of clinical Frey syndrome after parotidectomy is 40-50%. When objective tests are used (ISPH), the incidence is around 80%.

204 13) The use of an implant placed in the wound as a prevention barrier reduces the incidence of clinical Frey syndrome to 2-3%. When objective tests are used (ISPH) the incidence with e- PTFE is reduced to 10%.

14) The best Frey syndrome prevention barrier appears to be a non-resorbable implant.

15) Some of the implants used (mainly Ethisorb®, but also e-PTFE) result in a high incidence of parotid fistula. Therefore, the search for the best implant should continue.

16) The treatment of Frey syndrome by intradermal injection of botulinum toxin type A appears to be simple, effective, reliable, fast and devoid of major side effects. Further work is required to determine the minimal dosage per injection, the minimal inter-injection distance, as well as the long-term duration of the effect.

[1; 8; 41; 53; 68; 89; 104; 146; 151; 161; 176; 182; 200; 201; 202; 224; 251; 301; 329; 330; 336; 350; 365; 388; 389; 400; 405; 417; 423; 480; 544; 553]

205 APPENDIX 1

228 229 230 8. REFERENCES

1. Adie R. Gustatory sweating. Aust NZ J Surg 31:98-103,1968. 2. Adour KK, Swanson PJ. Facial paralysis in 403 consecutive patients: emphasis on treatment response in patients with Bell's palsy. Trans Am Acad Ophtalmol Otolaryngol 75:1284-1301,1971. 3. Adson AW, Ott WO. Preservation of the facial nerve in the radical treatment of parotid tumors. Arch Surg 6:739-746,1923. 4. Agnew WF, McCreery DB, Yuen TGH, Bullara LA. Histological and physiologic evaluation of electrically stimulated peripheral nerve: considerations for the selection of parameters. Ann Biomed Eng 17:39- 60,1989. 5. Ahlbom HE. Mucous and salivary-gland tumours; clinical study with special reference to radiotherapy, based on 254 cases treated at Radiumhemmet, Stockholm. Acta Radiol Suppl. 23:1-452,1935. 6. Alajmo E, Polli G, DeMeester W. Total parotidectomy - a routine treatment for parotid swellings? J Laryngol Otol 103:181-186,1989. 7. Allison GR, Rappaport I. Prevention of Frey's syndrome with SMAS interposition. Am J Surg 166:407- 410,1993. 8. Al-Naqeeb NI, Dashti H, Al-Muhanna AH, Behbehani A. Parotid gland tumours: a 15-year experience. J Royal Coll Surg (Edin) 37:89-93,1992. 9. Ammirati M, Spallone A, Ma J, Cheatham M, Becker D. An anatomicosurgical study of the temporal branch of the facial nerve. Neurosurg 33:1038-1044,1993. 10. Ananthakrishnan N, Parkash S. Parotid fistulas: a review. Brit J Surg 69:641-643,1982. 11. Appiani E. Abordaje para la parotidectomía y transplante muscular. Pren Med Argent 52:1242- 1243,1967. 12. Appiani E, Delfino MC. Plastic incisions for facial and neck tumors. Ann Plast Surg 13:335-352,1984. 13. Armistead PR, Smiddy FG, Frank HG. Simple enucleation and radiotherapy in the treatment of the pleomorphic salivary adenoma of the parotid gland. Brit J Surg 66:716-717,1979. 14. Arndt F, Bock M, Draf W. Parotidektomie: Funktionnelle und ästhetische Ergebnisse. HNO 39:271- 277,1991. 15. Baddour HM, Ripley JF, Cortez EA, McAnear JT, Steed DL, Tilson HB. Treatment of Frey's syndrome by an interpositional fascia graft: report of case. J Oral Surgery 38:778-781,1980. 16. Bailey H. The treatment of tumors of the parotid gland with special reference to total parotidectomy. Brit J Surg 28:337-346,1941. 17. Bailey H. Parotidectomy: indications and results. Br Med J 2:404-407,1947. 18. Bailey H. The surgical anatomy of the parotid gland. Br Med J 2:245-248,1948. 19. Baillarger M. Mémoire sur l'oblitération du canal de Sténon. Gazette Médicale de Paris 23:194- 197,1853. 20. Baker DC, Conley J. Avoiding facial nerve injuries in rhytidectomy. Plast Reconstr Surg 64:781- 795,1979. 21. Baker DC, Shaw WW, Conley J. Reconstruction of radical parotidectomy defects. Am J Surg 138:550- 554,1979. 22. Balfour HH, Bloom JE. The auriculotemporal syndrome beginning in infancy. J Pediatr 77:872-874,1970. 23. Bames H. Truth and fallacies of face peeling and face lifting. Med J Reconstr 26:86-98,1927.

205 24. Barbat JH. Complete removal of the parotid gland without injury of the facial nerve. California State J Med 14:115-116,1916. 25. Barton J, Slevin NJ, Gleave EN. Radiotherapy for pleomorphic adenoma of the parotid gland. Int J Rad Oncol Biol Phys 22:925-928,1992. 26. Basbaum CB. Induced hypothermia in peripheral nerve: electron microscopic and electrophysiological observations. J Neurocytol 2:171-187,1973. 27. Bassereau G, Touyeras A, Heitzmann P, Dauphin D. Exposition du nerf facial dans la region massetérine lors de chirurgie parotidienne. Ann Otolaryngol Chir Cervicofac 114:239-242,1997. 28. Bassoe PN. Auriculotemporal syndrome and other vasomotor disturbances about the head : "auriculotemporal syndrome" complicating diseases of parotid gland, angioedema of the brain. Med Clin NA 16:405-412,1932. 29. Batsakis JG, Chinn E, Regezzi JA, Repola DA. Pathology of head and neck tumors. II: Salivary glands. Head Neck Surg 1:167-180,1978. 30. Batsakis JG, Regezzi JA. Pathology of head and neck tumors. I: Salivary glands. Head Neck Surg 1:59- 68,1978. 31. Batsakis JG, Regezzi JA. Pathology of head and neck tumors. IV: Salivary glands. Head Neck Surg 1:340-349,1979. 32. Batsakis JG, Regezzi JA, Bloch D. Pathology of head and neck tumors. III: Salivary glands. Head Neck Surg 1:260-273,1979. 33. Beahrs OH, Adson MA. The surgical anatomy and technic of parotidectomy. Am J Surg 95:885- 896,1958. 34. Beahrs OH, L'Esperance BF. The facial nerve in parotid surgery. JAMA 162:261-263,1956. 35. Beahrs OH, Wollner LB, Carveth SW, Devine KD. Surgical management of parotid lesions. Review of seven hundred sixty cases. Arch Surg 80:890-904,1960. 36. Beck SA, Burks WA, Woody RC. Auriculotemporal syndrome seen clinically as food allergy. Pediatrics 83:601-603,1989. 37. Becker FF. The preauricular portion of the rhytidectomy incision. Arch Otolaryngol Head Neck Surg 120:166-171,1994. 38. Belding HH, Page DE. Parotid tumors. 15-year review of 98 cases. Am Surg 32:735-738,1966. 39. Belly E, Valentini V, Matteini C. Ruolo dello SMAS nella prevenzione della sindrome di Frey. Min Stomatol 45:569-574,1996. 40. Benedict EG, Meigs JV. Tumors of the parotid gland. A study of 225 cases with complete end-results in 80 cases. Surg Gynecol Obstet 51:626-647,1930. 41. Bentkover SH, Kozlicak BA, Girouard S. Outpatient parotidectomy at the Fallon clinic. Arch Otolaryngol Head Neck Surg 122:1049-1053,1996. 42. Bentley FH, Schlapp W. Experiments on the blood supply of nerves. J Physiol 102:62-85,1943. 43. Berard M. Extirpation de la parotide. Mémoires Observations Extraits 2:60-66,1823. 44. Bernstein L, Nielson RH. Surgical anatomy of the extraparotid distribution of the facial nerve. Arch Otolaryngol Head Neck Surg 110:177-183,1984. 45. Bigland B, Lippold OJ. The relation between force, velocity and integrated electrical activity in human muscles. J Physiol 123:214-224,1954. 46. Bjerkhoel A, Trobbe O. Frey's syndrome: treatment with botulinum toxin. J Laryngol Otol 111:839- 844,1997. 47. Bjorklund A, Eneroth CM. Management of parotid gland neoplasms. Am J Otolaryngol 1:155-167,1980. 48. Blair VP. Surgery and diseases of the mouth and jaws. St Louis, C.V. Mosby, 1912.

206 49. Blasi J, Chapman ER, Link E, Binz T, Yamasaki S, DeCamilli P, Sudhof T, Nieman H, Jahn R. Botulinum type A selectively cleaves the synaptic protein SNAP-25. Nature 365:160-163,1993. 50. Blunt MJ. The blood supply of the facial nerve. J Anat 88:520-526,1954. 51. Bonanno PC, Casson PR. Frey's syndrome: a preventable phenomenon. Plast Reconstr Surg 89: 452- 456,1992. 52. Borodic GE, Ferrante RJ, Pearce LB, Alderson K. Pharmacology and histology of the therapeutic application of botulinum toxin. In Therapy with botulinum toxin. Eds: Jankovic J, Hallett M. New York, Marcel Drekker Inc. pp:119-157,1994. 53. Borthne A, Kjellevold K, Kaalhus O, Vermund H. Salivary gland malignant neoplasms: treatment and prognosis. Int J Rad Oncol Biol Phys 12:747-754,1986. 54. Botkin S. Ueber die Reflexerscheinungen im Gebiete der Hautgefässe und ueber den reflectorischen Schweiss. Berliner Klinische Wochenschrift 7:81-83,1875. 55. Botman JWM, Jongkees LBW. The results of intratemporal treatment of facial palsy. Pract Otorhinolaryngol 17:80-100,1955. 56. Bouchet A, Cuilleret J. Anatomie topographique, descriptive et fonctionnelle. Villeurbanne, Simep Editions, 1971. 57. Brach JS, VanSwearingen J, Delitto A, Johnson PC. Impairement and disability in patients with facial neuromuscular dysfunction. Otolaryngol Head Neck Surg 117:315-321,1997. 58. Brainard D. Notes relating to the extirpation of the parotid gland. Chicago Med J 17:569-581,1860. 59. Brintall ES, Tidrick RT, Huffman WC. A simplified and rapid approach to parotidectomy. Arch Surg 71:331-336,1955. 60. Brodal A. Neurological anatomy in relation to clinical medicine. New York, Oxford University Press, 1981. 61. Bron LP, O'Brien CJ. Facial nerve function after parotidectomy. Arch Otolaryngol Head Neck Surg 123:1091-1096,1997. 62. Brown AMS, Wake MJC. Accidental full thickness burn of the ear lobe following division of the greater auricular nerve at parotidectomy. Br J Oral Maxillofac Surg 28:178-179,1990. 63. Brown JB, McDowell F, Fryer MP. Direct operative removal of benign mixed tumors of anlage origin in the parotid region. Surg Gynecol Obst 90:257-268,1950. 64. Brown JS, Ord RA. Preserving the greater auricular nerve in parotid surgery. Brit J Oral Maxillofac Surg 27:459-466,1989. 65. Brown MC, Hardman VJ. A reassessment of the accuracy of reinnervation of motoneurons following crushing or freezing of the sciatic and lumbar nerves of rats. Brain 110:695-705,1987. 66. Brown-Séquard CE. Production de sueur sous l'influence d'une excitation vive des nerfs du goût. Compte Rendu Société de Biologie 1:104,1850. 67. Bruesch SR. The distribution of myelinated afferent fibers in the branches of the cat's facial nerve. J Comp Neurol 81:169-191,1944. 68. Brusati R, Bozzetti A, Chiapasco M. Facial nerve and parotid surgery. J Cranio-Max-Fac Surg 15:278- 280,1987. 69. Buchmann C, Stringer SP, Mendenhall WM, Parsons JT, Jordan JR, Cassissi NJ. Pleomorphic adenoma: effect of tumor spill and inadequate resection on tumor recurrence. Laryngoscope 104:1231- 1234,1993. 70. Bugis SP, Young JEM, Archibald SD. Sternocleidomastoid flap following parotidectomy. Head Neck Surg 12:430-435,1990. 71. Burres S, Fisch U. The comparison of facial grading systems. Arch Otolaryngol Head Neck Surg 112:755-758,1986.

207 72. Burres SA. Facial biomechanics: the standards of normal. Laryngoscope 95:708-714,1985. 73. Burres SA. Objective grading of facial paralysis. Ann Otol Rhinol Laryngol 95:238-241,1986. 74. Burrows A, Douglas WR. Notes on the treatment of recurrent endothelioma of the parotid gland. Acta Radiol 6:217-220,1926. 75. Burton MJ, Brochwicz-Lewinski M. Lucja Frey and the auriculotemporal nerve syndrome. J R Soc Med 84:619-620,1991. 76. Bushara KO, Park DM, Jones JC, Schutta HS. Botulinum toxin - a possible new treatment for axillary hyperhydrosis. Clin Exp Derm 21:276-278,1996. 77. Butler ECB. The management of tumors of the parotid gland, based on 373 cases seen at the London Hospital between 1943-1958. Bull Soc Int Chir 23:10-16,1964. 78. Buxton RW, Maxwell JH, Franch AJ. Surgical treatment of epithelial tumors of the parotid gland. Surg Gynecol Obst 97:401-416,1953. 79. Byars LT. Preservation of the facial nerve in operations for benign conditions of the parotid area. Ann Surg 136:412-421,1952. 80. Cadenat FM. Extirpation des tumeurs bénignes de la parotide. Le digastrique, muscle directeur de l'opération et muscle correcteur de la paralysie faciale. J Chir 48:625-629,1936. 81. Cannoni M, Pech A, Scavennec C, Zanaret M, Dessi P. Les récidives des adénomes pléomorphes. Rev Laryngol 109:13-16,1988. 82. Carmichael R. An account of the extirpation of a tumour of the neck engaging the parotid gland. Trans Assoc Fell Licentiates Kings Queen Coll Physicians 101-113,1818. 83. Carr RJ, Bowerman JE. A review of tumours of the deep lobe of the parotid salivary gland. Br J Oral Maxillofac Surg 24:155-168,1986. 84. Carwardine T. Excision of the parotid gland with preservation of the facial nerve. Lancet ii:892,1907. 85. Casler JD, Conley JJ. Sternocleidomastoid muscle transfer and superficial musculoaponeurotic system plication in the prevention of Frey's syndrome. Laryngoscope 101:95-100,1991. 86. Cassady CL. A new concept in the treatment of Frey's syndrome: the use of interpositional dermal grafts. An experimental study in the dog. Laryngoscope 87:962-974,1977. 87. Casselman JW, Mancuso AA. Major salivary gland masses: comparison of MR imaging and CT. Radiology 165:183-189,1987. 88. Castro Correia P, Zani R. Surgical anatomy of the facial nerve, as related to ancillary operations in rhytidoplasty. Plast Reconstr Surg 52:549-552,1973. 89. Castro EB, Huvos AG, Strong EW, Foote FW. Tumors of the major salivary glands in children. Cancer 29:312-317,1972. 90. Chalk CH, Dyck PJ. Ischemic neuropathy. In Peripheral neuropathy. Eds: Dyck PJ, Thomas PK. Philadelphia, WB Saunders. pp:980-989,1993. 91. Champy M. Comblement des cavités de parotidectomie par lambeau musculo-périosté rétro-auriculaire. Ann Chir Plast 22:69-71,1977. 92. Chan MKM, McGuire LJ, King W, Li AKC, Lee JCK. Cytodiagnosis of 112 salivary gland lesions. Correlation with histologic and frozen section diagnosis. Acta Cytologica 36:353-363,1992. 93. Chan S, Gunn A. Conservative parotidectomy by the peripheral approach. Brit J Surg 68:405-407,1981. 94. Chevalier D, Loche V, Darras JA, Apko-Allavo J, Desaulty A, Piquet JJ. Les reprises et récidives d'adénome pléomorphe de la glande parotide. Ann Otolaryngol Chir Cervicofac 113:56-60,1996. 95. Christensen NR, Jacobsen SD. Parotidectomy. Preserving the posterior branch of the great auricular nerve. J Laryngol Otol 111:556-559,1997. 96. Clausen EG, Henley RB. The surgical management of mixed tumors of the parotid gland. California Medicine 68:366-369,1948.

208 97. Cocke EW. Preservation of the facial nerve in recurrent mixed tumor of the parotid gland. Laryngoscope 88:2032-2034,1978. 98. Coffield JA, Considine RV, Simpson LL. The site and mechanism of action of botulinum neurotoxin. In Therapy with botulinum toxin. Eds: Jankovic J, Hallett M. New York, Marcel Drekker Inc. pp:3-13,1994. 99. Cohen S. Purification of a nerve growth promoting protein from mouse salivary gland and its neurocytoxic antiserum. Proc Natl Acad Sci USA 46:302-311,1960. 100. Cohen S. Personal experience with an alternate incision for parotidectomy. J Otolaryngol 17:382- 384,1988. 101. Coker NJ, Fordice JO, Moore S. Correlation of the nerve excitability test and electroneurography in acute facial paralysis. Am J Otol 13:127-133,1992. 102. Coldwater KB. Surgical treatment of the auriculotemporal syndrome. Arch Surg 69:54-57,1954. 103. Coller FA, Yglesias L. Infections of the lip and face. Surg Gynecol Obst 60:277-289,1935. 104. Comoretto R, Barzan L. Benign parotid tumour enucleation - a reliable operation in selected cases. J Laryngol Otol 104:706-708,1990. 105. Conley J. Recurrent mixed tumors of the parotid. Laryngoscope 90:880,1980. 106. Conley J. Invited comment on "Plastic incisions for facial and neck tumors" by E. Appiani and M.C. Delfino. Ann Plastic Surg 13:351-352,1984. 107. Conley J, Clairmont AA. Facial nerve in recurrent benign pleomorphic adenoma. Arch Otolaryngol 105:247-251,1979. 108. Conley JJ. Search for and identification of the facial nerve. Laryngoscope 88:172,1978. 109. Croxson G, May M, Mester SJ. Grading facial nerve function: House-Brackmann versus Burres-Fisch methods. Am J Otol 11:240-246,1990. 110. Cuntiffe WJ, Johnson CE. Gustatory hyperhydrosis a complication of thyroidectomy. Brit J Derm 79:519- 526,1967. 111. Cushing H. The sensory distribution of the fifth cranial nerve. Bull Johns Hopkins Hosp 15:213- 232,1904. 112. Dale HH, Feldberg W. The chemical transmission of secretory impulses to the sweat glands of the cat. J Physiol 82:121-128,1934. 113. Danovan DT, Conley JJ. Capsular significance in parotid tumor surgery: reality and myths of lateral lobectomy. Laryngoscope 94:324-329,1984. 114. Dargent M, Duroux PE. Données anatomiques concernant la morphologie et certains rapports du facial intra-parotidien. Presse Med 54:523-524,1946. 115. Davis RA, Anson BJ, Budinger JM, Kurth RE. Surgical anatomy of the facial nerve and parotid gland based upon a study of 350 cervicofacial halves. Surg Gynecol Obst 102:385-412,1956. 116. Davis RS, Strunk RC. Auriculotemporal syndrome in children. Am J Dis Child 135:832-833,1981. 117. Dawson AK. Radiation therapy in recurrent pleomorphic adenoma of the parotid. Int J Rad Oncol Biol Phys 16:819-821,1989. 118. Dawson AK, Orr JA. Long-term results of local excision and radiotherapy in pleomorphic adenoma of the parotid. Int J Radiat Oncol Biol Phys 11:451-455,1985. 119. Debets JMH, Munting JDK. Parotidectomy for parotid tumours: a 19-year experience from the Netherlands. Brit J Surg 79:1159-1161,1992. 120. Delarue J. Les "tumeurs mixtes" plurifocales de la glande parotide. Ann Anat Pathol 1:34-58,1956. 121. Dendooven AM, Lissens M, Bruyninckx F, Vanhecke J. Electrical injury to peripheral nerves. Medica Physica 13:161-165,1990. 122. Dennis JM, Coker NJ. Electroneuronography. Adv Otorhinolaryngol 53:112-131,1997.

209 123. Denny-Brown D, Brenner C. Paralysis of nerve induced by direct pressure and by tourniquet. Arch Neurol Psychiatry 51:1-26,1944. 124. Depondt J, Koka VN, Nasser T, Portier F, Guedon C, Barry B, Gehanno P. Use of fibrin glue in parotidectomy closure. Laryngoscope 106:784-787,1996. 125. Dingman RO, Grabb WC. Surgical anatomy of the mandibular ramus of the facial nerve based on the dissection of 100 facial halves. Plast Reconstr Surg 29:266-272,1962. 126. Dole VP, Thaysen JH. Variation on the functional power of human sweat glands. J Exp Med 98:129- 145,1953. 127. Dorozynski A. The blessed Niels Stensen. Br Med J 297:1290,1988. 128. Drobik C, Laskawi R. Frey's syndrome: treatment with botulinum toxin. Acta Otolaryngol 115:459- 461,1995. 129. Drummond PD. Mechanisms of physiological gustatory sweating and flushing in the face. J Auton Nerv Syst 8:117-124,1995. 130. Drummond PD, Boyce GM, Lance JW. Postherpetic gustatory flushing and sweating. Ann Neurol 21:559-563,1987. 131. Dulguerov P. Parotidectomy complications: role of facial nerve monitoring. 100th Annual Meeting of the American Academy of Otolaryngology - Head and Neck Surgery, Washington DC, 1996. 132. Dulguerov P, Gysin C, Marchal F, Lehmann W. Meta-analysis of postparotidectomy facial nerve paralysis. Arch Otolaryngol Head Neck Surg Submitted. 133. Dulguerov P, Lehmann W. Apport du monitoring du nerf facial dans la chirurgie parotidienne. 90e Congrès de la Société Suisse d'Oto-Rhino-Laryngologie et Chirurgie Cervico-Faciale, Loeche-les-Bains, 1995. 134. Dulguerov P, Marchal F, Gysin C. Frey syndrome before Frey : The correct history. Laryngoscope In press. 135. Dulguerov P, Marchal F, Lehmann W. Parotidectomie et fonction faciale - résultats genevois 1994-1998. Méd & Hyg 56:1965-1970,1998. 136. Dulguerov P, Marchal F, Lehmann W. Syndrome de Frey - diagnostic, traitement et prévention. Méd & Hyg 56:1972-1980,1998. 137. Dulguerov P, Marchal F, Lehmann W. Postparotidectomy facial nerve paralysis : possible etiologic factors and results with routine facial nerve monitoring. Laryngoscope 109:754-762,1999. 138. Dulguerov P, Marchal F, Lehmann W. Frey syndrome - diagnosis, prevention, and treatment. Acta Otolaryngol Submitted. 139. Dulguerov P, Marchal F, Lehmann W. Gustatory sweating. N Engl J Med Submitted. 140. Dulguerov P, Marchal F, Wang D, Gysin C. Review of objective topographic facial nerve evaluation methods. Am J Otol In press. 141. Dulguerov P, Nasri S, Mickel R. Horner's syndrome in thyroid neoplams. ORL Related Spec Submitted. 142. Dulguerov P, Quinodoz D, Cosendai G, Piletta P, Lehmann W. Prevention of Frey syndrome during parotidectomy. Arch Otolaryngol Head Neck Surg In press. 143. Dulguerov P, Quinodoz D, Cosendai G, Piletta P, Lehmann W. Treatment of Frey syndrome with botulinum toxin. Otolaryngol Head Neck Surg In press. 144. Dulguerov P, Quinodoz D, Vaezi A, Cosendai G, Pilleta P, Lehmann W. New objective and quantitative tests for gustatory facial sweating. Acta Otolaryngol In press. 145. Dulguerov P, Wang D, Perneger TV, Lehmann W. Videomimicography: a new technique for objective evaluation of facial motor function. Ann Neurol Submitted. 146. Dunn EJ, Kent T, Hines J, Cohn I. Parotid neoplasms: a report of 250 cases and review of the literature. Ann Surg 184:500-506,1976.

210 147. Duphenix M. Observations sur les fistules du canal salivaire de Stenon. Sur une playe compliquée à la joue ou le canal salivaire fut déchiré. Mémoires Académie Royale de Chirurgie 3:431-439,1757. 148. Dupuy. Observations et expériences sur l'enlèvement des ganglions gutturaux des nerfs trisplanchniques, sur des chevaux. J Med Chir Pharm 37:340-366,1816. 149. Duval P. Conservation des rameaux supérieurs (fronto-palpébraux) du facial dans l'extirpation totale de la parotide en dehors du cancer. Rev Chir 49:132-136,1914. 150. Duval P, Redon H. Extirpation totale ou subtotale de la parotide avec conservation de la branche supérieure du facial dans les tumeurs dites mixtes de la glande. Résultats cliniques. J Chir 39:801- 808,1932. 151. Dykun RJ, Deitel M, Borowy ZJ, Jackson S. Treatment of parotid neoplasms. Can J Surg 23:14-19,1980. 152. Eddey HH. Recurrent parotid tumors. Aust NZ J Surg 39:356-361,1970. 153. Eickhorn R, Haverkampf K, Antoni H. Disturbances of neural conduction in isolated frog nerves following exposure to strong electric fields. Muscle & Nerve 9:313-318,1986. 154. Eisele DW, Johns ME. Salivary gland neoplasms. In Head and Neck Surgery - Otolaryngology. Eds: Bailey BJ. Philadelphia, WB Lippincott. pp:1125-1147,1993. 155. El-Naggar M, Rice B, Oswal V. Life-size photograph transparencies: a method for photographic detection and documentation of recovery from facial paralysis. J Laryngol Otol 109:748-750,1995. 156. Eneroth CM. Histological and clinical aspects of parotid tumors. Acta Otolaryngol (Stockh) Suppl 191:1964. 157. Eneroth CM. Mixed tumors of major salivary glands: prognostic role of capsular structure. Ann Otol 74:944-953,1965. 158. Eneroth CM. Salivary gland tumors in the parotid gland, submandibular gland, and palate region. Cancer 27:1415-1419,1971. 159. Eneroth CM, Hamberger CA. Principles of treatment of different types of parotid tumors. Laryngoscope 84:1732-1740,1974. 160. Eneroth CM, Zetterberg A. Malignant pleomorphic adenoma. A clinical and microspectrophotometric study. Acta Otolaryngol (Stockh) 77:426-432,1974. 161. Essaadi M, Raji A, Janah A, Detsouli M, Kadiri F, Laraqui NZ, Touhami M, Moukrim B, Benghalem H, Benchekroun Y. Les tumeurs de la parotide (à propos de 120 cas). Rev Laryngol (Bordeaux) 119:177- 182,1998. 162. Fardy MJ. Neurotic excoriations complicating superficial parotidectomy - a case report. Br J Oral Maxillofac Surg 31:41-42,1993. 163. Farisse J, Bonnoit J, Di Marino V, Lebreuil G, Guerinel G. A propos de l'anatomie chirurgicale de la région parotidienne. Bull Ass Anat 147:254-261,1970. 164. Farrell ML, Kalnins IK. Frey's syndrome following parotid surgery. Aust NZ J Surg 61:295-301,1991. 165. Faure JL. Etude anatomique sur l'extirpation de la parotide et la résection préliminaire du bord postérieur de la mâchoire. Gazette des Hôpitaux 68:353-362,1895. 166. Federspil PA, Federspil P, Schatzle W. Parotid pleomorphic adenomas and their recurrences. HNO 42:28-35,1994. 167. Fee WE Jr, Goffinet DR, Calcaterra TC. Recurrent mixed tumors of the parotid gland - results of surgical therapy. Laryngoscope 88:265-273,1978. 168. Fee WE Jr, Handen C. Parotid gland surgery using the Shaw hemostatic scalpel. Arch Otolaryngol 110:739-741,1984. 169. Feiwel M. Auriculotemporal syndrome. Proc R Soc Med 52:191,1959. 170. Ferreira JL, Maurino N, Michael E, Ratinoff M, Rubio E. Surgery of the parotid region: a new approach. J Oral Maxillofac Surg 48:803-807,1990.

211 171. Fields MJ, Peckitt NS. Facial nerve function index: a clinical measurement of facial nerve activity in patients with facial nerve palsies. Oral Surg Oral Med Oral Pathol 69:681-682,1990. 172. Fiess H, Klempa I, Hermanek P, Sulkowski U, Uhl W, Beger HG, MW B. Prophylaxis of complications after pancreatic surgery: results of a multicenter trial in Germany. Digestion 55:Suppl 1: 35-40,1994. 173. Finochietto R, Marino H, Zavaleta D. Parotida - su extirpation con conservacion total del nervio facial. Revista Medica Municipal (Rio de Janeiro) 2:790-803,1941. 174. Fisch U. Surgery for Bell's palsy. Arch Otolaryngol 107:1-11,1981. 175. Fitzgibbon GM. Discussion on the surgical treatment of parotid tumours. Proc R Soc Med 46:1016- 1018,1953. 176. Fitzpatrick PJ, Black KM. Salivary gland tumors. J Otolaryngol 14:296-300,1985. 177. Foley JO, Dubois FS. An experimental study of the facial nerve. J Comp Neurol 79:79-105,1943. 178. Foote FW, Frazell EL. Tumors of the major salivary glands. Cancer 6:1065-1133,1953. 179. Ford FR, Woodhall B. Phenomena due to misdirection of regenerating fibers of cranial, spinal, and autonomic nerves: Clinical observations. Arch Surg 36:480-496,1938. 180. Fox RH, Goldsmith R, Kidd DJ. Cutaneous vasomotor control in the human head, neck, and upper chest. J Physiol 161:298-312,1962. 181. Franz D, Iggo A. Conduction failure in myelinated and non-myelinated axons at low temperatures. J Physiol 199:319-345,1968. 182. Frazell EL. Clinical aspects of tumors of the major salivary glands. Cancer 7:637-659,1954. 183. Freedberg AS, Shaw RS, McManus MJ. The auriculotemporal syndrome, a clinical and pharmacologic study. J Clin Invest 27:669-676,1948. 184. Freedman A. Facial sweating after food in diabetes. Br Med J 3:291,1973. 185. Frey L. Le syndrome du nerf auriculo-temporal. Rev Neurol 2:92-104,1923. 186. Frey M, Jenny A, Giovanoli P, Stussi E. Development of a new documentation system for facial movements as a basis for the international registry for neuromuscular reconstruction of the face. Plast Reconstr Surg 93:1334-1349,1994. 187. Furnas DW. Landmarks for the trunk and the temporofacial division of the facial nerve. Brit J Surg 52:694-696,1965. 188. Gardner WJ, McCubbin JW. Auriculotemporal syndrome: gustatory sweating due to misdirection of regenerated nerve fibers. JAMA 160:272-277,1956. 189. Garrett JR. Changing attitudes on salivary secretion - a short history on spit. Proc R Soc Med 68:553- 560,1975. 190. Gasser R. The early development of the parotid gland around the facial nerve and its branches in man. Anat Rec 167:63-78,1970. 191. Gates GA. Nerve excitability testing: technical pitfalls and threshold norms using absolute values. Laryngoscope 103:379-385,1993. 192. Gaughran GRL. The parotid compartment. Ann Otol 70:31-51,1961. 193. Glaister DH, Hearnshaw JR, Heffron PF, Patey DH. The mechanism of post-parotidectomy gustatory sweating (The auriculo-temporal syndrome). Brit Med J 2:942-946,1958. 194. Gleave EN, Whittaker JS, Nicholson A. Salivary tumors - experience over thirty years. Clin Otolaryngol 4:247-257,1979. 195. Goatcher PD. The Auriculo-Temporal syndrome. Br Med J 1:1233-1234,1954. 196. Gola R, Chossegros C, Carreau P. Anatomie chirurgicale de la région parotidienne. Rev Stomatol Chir Maxillofac 95:395-410,1994. 197. Golding-Wood PH. Tympanic neurectomy. J Laryngol Otol 76:683-693,1962.

212 198. Goodman RS. Frey's syndrome: secondary to condylar fracture. Laryngoscope 96:1397-1398,1986. 199. Goodnight JW, Dulguerov P, Berke GS, Lesavoy M, Hoffman LE. Correlation of compound action potential and EMG with facial muscle tension. Otolaryngol Head Neck Surg 112:279-290,1995. 200. Goras J, Politis A, Besbeas S, Georgacas G, Papamatheakis J. Neoplasms of the parotid gland. Review of 185 operated cases. Int Surg 58:178-181,1973. 201. Gordon AB, Fiddien RV. Frey's syndrome after parotid surgery. Am J Surg 132:54-58,1976. 202. Gore DO, Annamunthodo H, Harland A. Tumors of salivary origin. Surg Gynecol Obst 119:1290- 1296,1961. 203. Gosain AK. Surgical anatomy of the facial nerve. Clin Plast Surg 22:241-251,1995. 204. Gosain AK, Yousif NJ, Madiedo G, Larson DL, Matloub HS, Sanger JR. Surgical anatomy of the SMAS: a reinvestigation. Plast Reconstr Surg 92:1254-1263,1993. 205. Grage TB, Lober PH, Shahon DB. Benign tumors of the major salivary glands. Surgery 50:625- 633,1961. 206. Graney DO. Developmental anatomy. In Otolaryngology - Head and Neck Surgery. Eds: Cummings CW. St Louis, Mosby. pp:1517-1529,1986. 207. Grégoire R. Le nerf facial et la parotide. J Anat Physiol (Paris) 48:437-447,1912. 208. Grewal R, Xu J, Sotereanos DG, Woo SLY. Biomechanical properties of peripheral nerves. Hand Clin 12:195-204,1996. 209. Grodinsky M, Holyoke EA. The fascia and fascial spaces of the head and neck and adjacent regions. Am J Anat 63:367-408,1938. 210. Guerrerosantos J, Dicksheet S, Guillen C, Andini N. Hidden incision in surgery of parotid, submandibular, cervical, and cheek benign tumors. Ann Plast Surg 9:402-408,1982. 211. Guerrier B, Mustin V, Makieff M, Maurice N, Crampette L. Récidives et reprises des adénomes pléomorphes de la parotide. J Fr ORL 42:256-261,1993. 212. Guerrier Y, Mounier-Kuhn P. Histoire des maladies de l'oreille, du nez et de la gorge. Les grandes étapes de l'Oto-Rhino-Laryngologie. Paris, Roger Dacosta, 1980. 213. Gunn A, Parott NR. Parotid tumours: a review of parotid tumour surgery in the Northern Regional Health Authority of the United Kingdom 1978-1982. Br J Surg 75:1144-1146,1988. 214. Gunnel F. On the behavior of the capsule in tumours of the parotid gland. Arch Chr Kehlkopfheilk 178:291-300,1961. 215. Guth L. Regeneration in the mammalian peripheral nervous system. Physiol Rev 36:441-478,1956. 216. Gutierrez A. Tumores de la glandula parotida: Su extirpacion. Rev Cirugia 3:23,1924. 217. Guttman L. A demonstration of the study of sweat secretion by the quinazarin method. Proc Roy Soc Med 35:77-78,1941. 218. Haenggeli CA, Dulguerov P, Richter M, Lehmann W. Complications of operative ENG facial nerve monitoring. Am J Otol In press. 219. Haftek J. Stretch injury of peripheral nerve: acute effects of stretching on rabbit nerve. J Bone Joint Surg 52B:354-365,1970. 220. Hagan WE, Anderson JR. Rhytidectomy techniques utilized for benign parotid surgery. Laryngoscope 98:711-715,1980. 221. Hanna DC, Dickason WL, Richardson GS, Gaisford JC. Management of recurrent salivary gland tumors. Am J Surg 132:453-458,1976. 222. Harada T, Inoue T, Harashina T, Hatoko M, Ueda K. Dermis-fat graft after parotidectomy to prevent Frey's syndrome and the concave deformity. Ann Plast Surg 31:450-452,1993. 223. Harrison K, Donaldson I. Frey's syndrome. J R Soc Med 72:503-508,1979.

213 224. Haw CS. Pleomorphic adenoma of the parotid gland: A review of results of treatment. J R Coll Surg (Edinburgh) 20:25-29,1975. 225. Hawe MJG, Millar Bell D. Superficial parotidectomy for pleomorphic adenoma. Clin Oncol 8:207- 213,1982. 226. Haxton HA. Gustatory sweating. Brain 71:16-25,1948. 227. Hayes PC, Tulloch JA. An usual case of gustatory sweating. Postgrad Med J 58:656-657,1982. 228. Hays LL. The Frey syndrome: a review and double blind evaluation of the topical use of a new anticholinergic agent. Laryngoscope 88:1796-1824,1978. 229. Hays LL, Novack AJ, Worsham JC. The Frey syndrome: a simple effective treatment. Otolaryngol Head Neck Surg 90:419-425,1982. 230. Heeneman H. Identification of the facial nerve in parotid surgery. Can J Otolaryngol 4:145-151,1975. 231. Helling TD, Neely JG. Validation of objective measures for facial paralysis. Laryngoscope 107:1345- 1349,1997. 232. Helmus C. Subtotal parotidectomy: a 10-years review (1985-1994). Laryngoscope 107:1024- 1027,1997. 233. Hemenway WG. Gustatory sweating and flushing. The auriculo-temporal syndrome - Frey's syndrome. Laryngoscope 70:84-90,1960. 234. Herrmann F, Prose PH, Sulzberger MB. Studies on sweating. IV. A new quantitative method of assaying sweat-delivery to circumscribed areas of the skin surface. J Invest Dermatol 17:241-249,1951. 235. Hinderer UT. Prevention of insatisfactory scarring. Clin Plast Surg 4:199-205,1977. 236. Hirschenfang S, Goldberg MJ, Benton JG. Psychological aspects of patients with facial paralysis. Dis Nerv Syst 30:257-261,1969. 237. Hogg SP, Kratz RC. Surgical exposure of the facial nerve. Arch Otolaryngol 67:560-561,1958. 238. Holman ML, Neely JG, Joaquin AH. Computer-assisted clinimetric tool for the study of facial expression: a pilot validity study. Laryngoscope 106:135-138,1996. 239. Holt JJ. The stylomastoid area: anatomic-histologic study and surgical approach. Laryngoscope 106:396-400,1996. 240. Horner F. Ueber eine Form von Ptosis. Klin Monatsbl Augenheilk 7:193-198,1869. 241. Houck JW. Tumors of the salivary glands. II. A study of 48 cases with the presentation of a new clinical classification. Surgery 6:563-584,1939. 242. House JW. Facial nerve grading systems. Laryngoscope 93:1056-1069,1983. 243. House JW, Brackmann DE. Facial nerve grading system. Otolaryngol Head Neck Surg 93:146-147,1985. 244. Huber A, Schmid S, Fisch U. Pleomorphes Adenom der Glandula parotis. HNO 42:553-558,1994. 245. Hughes RG, Lyons TJ. Parotid tumours. Lancet 1:1080,1982. 246. Huttenbrink KB. Die Therapie des gustatorischen Schwitzens nach Parotidectomie. Laryngo Rhinol Otol 65:135-137,1986. 247. Hybbinette S. Contribution à l'étude des tumeurs bénignes de la parotide et à leur traitement radio- chirurgical. Acta Chir Scan 77:19-49,1936. 248. Iizuka K, Ishikawa K. Surgical techniques for benign parotid tumors: segmental resection vs extracapsular lumpectomy. Acta Otolaryngol (Stockh) Suppl 537:75-81,1998. 249. Isogai N, Kamiishi H. Application of medical thermography to the diagnosis of Frey syndrome. Head & Neck 19:143-147,1997. 250. Isono M, Murata K, Tanaka H, Kawamoto M, Azuma H. An objective evaluation method for facial mimic motion. Otolaryngol Head Neck Surg 114:27-31,1996.

214 251. Jackson GL, Luna MA, Beahrs RM. Results of surgery alone and surgery combined with postoperative radiotherapy in the treatment of cancer of the parotid gland. Am J Surg 146:497-500,1983. 252. Janes RM. The treatment of tumours of the salivary glands by radical excision. Can Med Assoc J 43:554-559,1940. 253. Janes RM. Tumours of the parotid gland. Ann R Coll Surg (England) 21:1-20,1957. 254. Janka HU, Standl E, Menhert H. Clonidine effect on diabetic gustatory sweating. Br Med J 91:130,1979. 255. Jankovic J, Hallet M. Therapy with botulinum toxin. New York, Marcel Dekker Inc., 1994. 256. Jefferson G, Platt H. The parotid gland. Anat Anz 41:81-89,1912. 257. Johnson IJM, Birchall JP. Bilateral auriculotemporal syndrome in childhood. Int J Ped Otorhinolaryngol 32:83-86,1995. 258. Johnson PC, Brown H, Kuzon WM, Balliet R, Garrison JL, Campbell J. Simultaneous quantitation of facial movements: the maximal static repsonse assay of facial nerve function. Ann Plastic Surg 32:171- 179,1994. 259. Jones KR. Anatomy of the neck. In The Neck. Diagnosis and surgery. Eds: Shockley WW, Pillsbury HC III. St Louis, Mosby. pp:3-17,1994. 260. Jones ME. The salivary glands: anatomy and embryology. Otolaryngol Clin North Am 10:261-271,1977. 261. Jones ME. Parotid cancer - a rational basis for treatment. Head Neck Surg 3:132-141,1980. 262. Jones ME, Kaplan MJ. Surgical therapy of tumors of the salivary glands. In Comprehensive management of Head and Neck tumors. Eds: Thawley SE, Panje WR. Philadelphia, WB Saunders. pp:1104-1138,1987. 263. Joseph J. Hängewangenplastik (Melomioplastik). Dtsch Med Wochenschr 47:287-291,1921. 264. Jost G, Legent F, Baupelot S. Le comblement des dépressions résiduelles après parotidectomie par un lambeau du sterno-cléîdo-mastoidien: Utilisation de l'incision de Lotte. Ann Oto-Laryngol 85:357- 360,1968. 265. Jost G, Levet Y. Parotid fascia and face lifting: a critical evaluation of the SMAS concept. Plast Reconstr Surg 74:42-51,1984. 266. Juvara E: Anatomie de la région ptérygomaxillaire. Paris1870. Cited by L.Testut and M. Grodinsky. 267. Kane WJ, McCaffrey TV, Olsen KD, Lewis JE. Primary parotid malignancies. A clinical and pathologic review. Arch Otolaryngol Head Neck Surg 117:307-315,1991. 268. Katz AD, Catalano P. The clinical significance of the various anastomotic branches of the facial nerve. Arch Otolaryngol Head Neck Surg 113:959-962,1987. 269. Kennedy WR, Sakuta M, Sutherland D, Goetz FC. Quantitation of sweating deficit in diabetes mellitus. Ann Neurol 15:482,1984. 270. Kennel P, Herman D, Piller P, Stierle JL, Conraux C. Les tumeurs de la loge parotidienne. A propos de 203 explorations chirurgicales. Rev Laryngol (Bordeaux) 113:105-110,1992. 271. Kerl J, Staubesand J. Vascular thermocoagulation - perivascular nerve lesions. An ultrastructural report on the choice between monopolar and bipolar electrocoagulation. Acta Neurochir 92:123-131,1988. 272. Kidd HA. Complete excision of the parotid gland with preservation of the facial nerve. Br Med J i:989- 991,1950. 273. Kidd HA. Total parotidectomy with preservation of facial nerve. Brit Med J ii:1309-1310,1955. 274. Kirklin JW, McDonald JR, Harrington SW, New GB. Parotid tumors. Histopathology, clinical behaviour, and end results. Surg Gynecol Obst 92:721-733,1952. 275. Kleinsasser O. Behavior and adequate treatment of epithelial tumors of the parotid gland. In Salivary glands and the facial nerve. Eds: Conley J. Stuttgard, G. Thieme. pp:15-23,1975.

215 276. Klopp CT, Winship T. Treatment of mixed tumors of the parotid gland by subtotal parotidectomy. Arch Surg 61:477-486,1950. 277. Kopuz C, Ilgi S, Yavuz S, Onderglu S. Morphology of the retromandibular vein in relation to the facial nerve in the parotid gland. Acta Anat 66-68,1995. 278. Kornblut AD, Westphal P, Miehlke A. The effectiveness of a sternomastoid muscle flap in preventing post-parotidectomy occurrence of the Frey syndrome. Acta Otolaryngol 77:368-373,1974. 279. Kornblut AD, Westphal P, Miehlke A. A re-evaluation of the Frey syndrome following parotid surgery. Arch Otolaryngol 103:258-261,1977. 280. Kozma C, Gabriel S. Gustatory flushing syndrome. A pediatric case and review of the literature. Clin Pediatr (Phila) 32:629-631,1993. 281. Krolls SO, Boyers RC. Mixed tumors of salivary glands: long term follow-up. Cancer 30:276-281,1972. 282. Krychatalsky B, Weinberg S. An assessment for auriculotemporal syndrome following temporomandibular joint surgery through the preauricular approach. J Oral Maxillofac Surg 47:3- 6,1989. 283. Kun Z, Dao-Yi Q, Li-Min W. Functional superficial parotidectomy. J Oral Maxillofac Surg 52:1038- 1041,1994. 284. Laage-Hellman J-E. Gustatory sweating and flushing after conservative parotidectomy. Acta Otolaryngol 48:234-252,1957. 285. Laage-Hellman J-E. Gustatory sweating and flushing. Aetiological implications of response of latent period and mode of development after parotidectomy. Acta Otolaryngol 49:306-314,1958. 286. Laage-Hellman J-E. Treatment of gustatory sweating and flushing. Acta Otolaryngol 49:132-143,1958. 287. Laage-Hellman J-E. Gustatory sweating and flushing. Aetiological implications of response of separate sweat glands to various stimuli. Acta Otolaryngol 49:363-374,1959. 288. Laccourreye H, Laccourreye O, Cauchois R, Jouffre V, Menard M, Brasnu D. Total conservative parotidectomy for primary benign pleomorphic adenoma of the parotid gland: a 25-year experience with 229 patients. Laryngoscope 104:1487-1494,1994. 289. Laccourreye O, Bernard D, Lacharrière O, Bazin R, Brasnu D. Frey's syndrome analysis with biosensor: A preliminary study. Arch Otolaryngol Head Neck Surg 119:940-944,1993. 290. Laccourreye O, Bonan B, Brasnu D, Laccourreye H. Treatment of Frey's syndrome with topical 2% Diphenanil Methylsulfate (Prantal): a double-blind evaluation of 15 patients. Laryngoscope 100:651- 653,1990. 291. Laccourreye O, Brasnu D, Jouffre V, Cauchois R, Naudo P, Laccourreye H. Dysfonctionnement du nerf facial après parotidectomie totale conservatrice de première intention pour adénome pléomorphe. Ann Otolaryngol Chir Cervicofac 112:63-68,1995. 292. Laccourreye O, Gutierrez-Fonseca R, Garcia D, Brasnu D, Bonan B. Recurrent gustatory sweating (Frey syndrome) after intracutaneous injection of botulinum toxin type A. Arch Otolaryngol Head Neck Surg 125:283-286,1999. 293. Laccourreye O, Muscatello L, Gutierrez-Fonseca R, Seckin S, Brasnu D, Bonan B. Syndrome de Frex sévère post-parotidectomie: traitement par la neuro-toxine botulinique de type A. Ann Otolaryngol Chir Cervicofac 116:137-142,1999. 294. Laccourreye O, Muscatelo L, Nuade C, Bonan B, Brasnu D. Botulinum toxin type A for Frey syndrome: a preliminary prospective study. Ann Otol Rhinol Laryngol 107:52-55,1998. 295. Lachard J, Vitton J, Gola R. A propos des parotidectomies pour lésions inflammatoires. Rev Stomatol (Paris) 73:235-238,1972. 296. Lacomme Y. Les récidives des adénomes pléomorphes de la parotide. Rev Laryngol (Bordeaux) 109:17- 24,1988.

216 297. Laing MR, McKerrow WS. Intraparotid anatomy of the facial nerve and the retromandibular vein. Brit J Surg 75:310-312,1988. 298. Lam CM, Wong SY. Use of somatostatin analog in the management of traumatic parotid fistula. Surgery 119:481-482,1996. 299. Lam KM, Wei W, Ho HC, Ho CM. Whole organ sectioning of mixed parotid tumors. Am J Surg 160:377- 380,1990. 300. Langeskiöld A. Gustatory local hyperhydrosis following injuries of the parotid region. Acta Chir Scand 93:294-306,1946. 301. Lanier VC, McSwain B, Rosenfeld L. Mixed tumors of salivary glands: a 44 year study. South Med J 65:1485-1488,1972. 302. Laskawi R, Drobik C, Schönebeck C. Up-to-date report of Botulinum toxin type A treatment in patients with gustatory sweating (Frey's syndrome). Laryngoscope 108:381-384,1998. 303. Lathrop FD. Management of the facial nerve during operations on the parotid gland. Ann Otol 72:780- 781,1963. 304. Lawson HH. Value of the tympanomastoid fissure in the exposure of the facial nerve. Brit J Surg 75:309,1988. 305. Lawson HH. Capsular penetration and perforation in pleomorphic adenoma of the parotid salivary gland. Brit J Surg 76:594-596,1989. 306. Layfield LJ, Tan P, Glasgow BJ. Fine-needle aspiration of salivary gland lesions. Arch Pathol Lab Med 111:346-353,1987. 307. Le Quesne PM. Neuropathy due to drugs. In Peripheral neuropathy. Eds: Dyck PJ, Thomas PK. Philadelphia, WB Saunders. pp:1571-1581,1993. 308. Lee TS. Physiological gustatory sweating in a warm climate. J Physiol 124:528-542,1954. 309. Lehmann W. Personal communication. 1994. 310. Lempert J. Tympanosympathectomy: A technique for the relief of tinnitus aurium. Arch Otolaryngol 43:199-212,1946. 311. Lennox B, Clarke JA, Drake F, Ewen SWB. Incidence of salivary gland tumors in Scotland: accuracy of national records. Brit Med J i:687-689,1978. 312. Lenormant C. Extirpation de la glande parotide. In Chirurgie de la tête et du cou. Eds: . Paris, Masson. pp:198-204,1919. 313. Leriche R. Behandlung der permanenten parotis Fisteln durch die Entnervung der Speicheldrüse Ausreissen des N. auriculotemporalis. Zentralblatt Chir 41:754-755,1914. 314. Leroux-Robert J. Notre expérience des tumeurs dites "mixtes" (ou épithéliomas à stroma remanié) de la parotide. Cahiers d'ORL 7:361-371,1972. 315. Leverstein H, van der Wal JE, Tiwari RM, van der Waal I, Snow GB. Surgical management of 246 previously untreated pleomorphic adenoma of the parotid gland. Brit J Surg 84:399-403,1997. 316. Levi-Montalchini R. Developmental neurobiology and natural history of nerve growth factor. Ann Rev Neurosci 5:341-362,1982. 317. Liebman EP, Webster RC, Gaul JR, Griffin T. The marginal mandibular nerve in rhytidectomy and liposuction surgery. Arch Otolaryngol Head Neck Surg 114:179-181,1988. 318. Linder TE. Personal communication. 1998. 319. Linder TE, Huber A, Schmid S. Frey's syndrome after parotidectomy: a retrospective and prospective analysis. Laryngoscope 107:1496-1501,1997. 320. List CF, Peet MM. Sweat secretion in man: IV. Sweat secretion of the face and its disturbance. Arch Neurol Psych 40:443-470,1938.

217 321. Longuercio C, De Sio L, Romano M. Effect of somatostatin on salivary secretion in man. Digestion 36:91-95,1987. 322. Lore JM Jr. An Atlas of Head and Neck Surgery. Philadelphia, WB Saunders, 1988. 323. Low PA. Quantitation of autonomic function. In Peripheral neuropathy. Eds: Dyck PJ, Thomas PK, Griffin JW, Low PA, Podulso JF. Philadelphia, W.B. Saunders Company. pp:729-745,1993. 324. Lundborg G, Rydevik B, Manthorpe M, Varon S, Lewis J. Peripheral nerve: The physiology of injury and repair. In Injury and repair of musculoskeletal soft tissue. Eds: Woo SLY, Buckwalter JA. Park Ridge, American Academy of Orthopedic Surgeons. pp:297-352,1987. 325. Lyle MF. Surgical consideration of parotid tumors. Am J Surg 91:332-338,1956. 326. Maas CS, Gnepp DR, Bumpous J. Expanded polytetrafluoroethylene (Gore-Tex soft-tissue patch) in facial augmentation. Arch Otolaryngol Head Neck Surg 119:1008-1014,1993. 327. MacGregor FC. Facial dysfigurement: problems and management of social interaction and implication for mental health. Aesthetic Plast Surg 14:249-257,1990. 328. Mailander JC. Hereditary gustatory sweating. JAMA 201:121-122,1967. 329. Maimaris CV, Ball MJ. Treatment of parotid gland tumours by conservative parotidectomy. Brit J Surg 73:897,1986. 330. Maran AGD, Mackenzie IJ, Stanley RE. Recurrent pleomorphic adenomas of the parotid gland. Arch Otolaryngol 110:167-171,1984. 331. Marchac D, Sandor G. Face lifts and sprayed fibrin glue: an outcome analysis of 200 patients. Brit J Plast Surg 47:306-309,1994. 332. Marchal F, Dulguerov P, Becker M, Lehmann W. Interventional Sialendoscopy. In Minimally Invasive Surgery of the Head, Neck and Cranial Base. Eds: Wackym PA, Rice DH, Schaefer SD. Philadelphia, Lippincott, Williams & Wilkins. ,In press. 333. Marchal F, Dulguerov P, Lehmann W. Sialendoscopy. N Engl J Med In press. 334. Marshall SF, Miles GO. Tumors of the submaxillary and parotid gland areas. Surg Clin North Am 33:501- 516,1947. 335. Martin H. The operative removal of tumors of the parotid salivary gland. Surgery 31:670-682,1952. 336. Martis C. Parotid benign tumors: comments on surgical treatment of 263 cases. Int J Oral Surg 12:211- 220,1983. 337. Matthews B, Robertson PP. The course of post-ganglionic sympathetic fibers distributed with the trigeminal nerve in the cat. J Physiol 303:391-401,1980. 338. May JS, McGuirt WF. Frey's syndrome: treatment with topical glycopyrrolate. Head Neck 11:85- 89,1989. 339. May M, Blumenthal F, Taylor FH. Bell's palsy: surgery based on prognostic indicators and results. Laryngoscope 91:2092-2103,1981. 340. Maynard JD. Management of pleomorphic adenoma. Br J Surg 75:305-308,1988. 341. McClellan J. Case of extirpation of the parotid gland. Am J Med Sci 7:499-500,1844. 342. McCreery DB, Agnew WF, Yuen TGH, Bullara LA. Relationship between stimulus amplitude, stimulus frequency and neural damage during electrical stimulation of sciatic nerve of cat. Med Biol Eng Comput 33:426-429,1995. 343. McEvedy BV, Ross WM. The treatment of mixed parotid tumours by enucleation and radiotherapy. Brit J Surg 63:341-342,1976. 344. McFarland J. Ninety tumors of the parotid region, in all of which the postoperative history was traced. Am J Med Sci 172:804-820,1926. 345. McFarland J. Adenoma of the salivary glands; with report of a possible case. Am J Med Sci 174:362,1927.

218 346. McFarland J. Tumors of the parotid region: studies of one hundred and thirty-five cases. Surg Gynecol Obst 57:104-114,1933. 347. McFarland J. Three hundred mixed tumors of the salivary glands, of which sixty-nine recurred. Surg Gynecol Obst 63:457-468,1936. 348. McFarland J. Mysterious mixed tumors of salivary gland. Surg Gynecol Obst 76:23-34,1943. 349. McGregor AD, Burgoyne M, Tan KC. Recurrent pleomorphic salivary adenoma - the relevance of age at first presentation. Br J Plast Surg 41:177-181,1988. 350. McGurk FM, Main JHP, Orr JA. Adenolymphoma of the parotid gland. Brit J Surg 57:321-325,1970. 351. McKenzie J. The parotid gland in relation to the facial nerve. J Anat 82:183-186,1948. 352. McKinney P, Katrana DJ. Prevention of injury to the great auricular nerve during rhytidectomy. Plast Reconstr Surg 66:675-679,1980. 353. McNealy RW, McCallister JWM. Parotid gland tumors and their surgical management. Am J Surg 83:648-652,1952. 354. McWhorter J. The relations of the superficial and deep lobes of the parotid gland to the ducts and to the facial nerve. Anat Rec 12:449-454,1917. 355. Mehle ME, Kraus DH, Wood BG, Benninger MS, Eliachar I, Levine HL, Tucker HM, Lavertu P. Facial nerve morbidity following parotid surgery for benign disease: The Cleveland Clinic Foundation experience. Laryngoscope 103:386-388,1993. 356. Meier-Gallati V, Scriba H, Fisch U. Objective scaling of facial nerve function based on area analysis (OSCAR). Otolaryngol Head Neck Surg 118:545-550,1998. 357. Mellinkoff SM, Mellinkoff J. Gustatory hyperhydrosis of the left knee. JAMA 142:901-902,1950. 358. Micheli - Pellegrini V. Storia delle ghiandole salivari. Nuovo Arch Ital Otol 1:304-345,1973. 359. Milner-Brown HS, Stein RB. The relation between the surface electromyogram and muscular force. J Physiol 246:549-569,1975. 360. Minor V. Ein neues Verfahren zu der klinischen Untersuchung der Schweissabsonderung. Dtsch Z Nervenheilkd 101:302-308,1927. 361. Mitz V, Peyronie M. The superficial musculoaponeurotic system (SMAS) in the parotid and cheek area. Plast Reconstr Surg 58:80-88,1976. 362. Monte Pucelli F. Exposure of the facial nerve in parotid surgery: a study of the use of the tympanomastoid suture as a landmark. Am Surg 29:657-659,1963. 363. Montecucco C, Schiavo G, Tugnoli V, De Grandis D. Botulinum neurotoxins: mechanism of action and therapeutic applications. Mol Med Today 2:418-424,1996. 364. Morfit HM, Kramish D. Auriculotemporal syndrome (Frey's syndrome) following surgery of parotid tumors. Am J Surg 102:777-780,1961. 365. Morgan MN, MacKenzie DH. Tumours of salivary glands. A review of 204 cases with 5-year follow-up. Brit J Surg 55:284-288,1968. 366. Mott V. Case of extirpation of the parotid gland. Am J Med Sci 10:17-20,1832. 367. Mra Z, Komisar A, Blaugrund SM. Functional facial nerve weakness after surgery for benign parotid tumors: a multivariate statistical analysis. Head Neck 15:147-152,1993. 368. Murray JG, Thompson JW. Collateral sprouting in response to injury of the autonomic nervous system and its consequences. Br Med Bull 13:213-219,1957. 369. Murty GE, Diver JP, O'Donoghue GM, Bradley PJ. The Nottingham system: objective assessment of facial nerve function in the clinic. Otolaryngol Head Neck Surg 110:156-161,1994. 370. Myers EN, Conley J. Gustatory sweating after radical neck dissection. Arch Otolaryngol 91:534- 542,1970.

219 371. Myssiorek D, Ruah CB, Hybels RL. Recurrent pleomorphic adenomas of the parotid gland. Head & Neck 12:332-336,1990. 372. Naiem F, Forsberg MI, Waisman J, Coulson MF. Mixed tumors of the parotid gland. Growth pattern and recurrence. Arch Pathol Lab Med 100:271-275,1976. 373. Naumann M, Zellner M, Toyka KV, Reiners K. Treatment of gustatory sweating with botulinum toxin. Ann Neurol 42:973-975,1997. 374. Nawwar AM, Sherif MF, Barakat NG. Intrumental forceps for measurement of nerve compression forces. J Biomech Eng 117:53-58,1995. 375. Needles W. The auriculotemporal syndrome. With a suggestion regarding therapy. Arch Neurol Psychiatr 35:357-360,1936. 376. Neel HB. Implants of Gore-Tex. Comparisons with Teflon-coated polytetrafluoroethylene carbon and porous polyethylene implants. Arch Otolaryngol 109:427-433,1983. 377. Neely JG, Cheung JY, Wood M, Byers J, Rogerson A. Computerized quantitative dynamic analysis of facial motion in the paralyzed and synkinetic face. Am J Otol 13:97-107,1992. 378. Neely JG, Jekel JF, Cheung JY. Variations in maximum amplitude of facial expressions between and within normal subjects. Otolaryngol Head Neck Surg 110:60-63,1994. 379. Neely JG, Joaquin AH, Kohn LA, Cheung JY. Quantitative assessment of the variation within grades of facial paralysis. Laryngoscope 106:438-442,1996. 380. Neely JG, Neufeld PS. Defining functional limitation, disability, and social limitations in patients with facial paresis: initial pilot questionnaire. Am J Otol 17:340-342,1996. 381. Neisse R. Ueber den Einschluss von Parotislappchen in Lymphknoten. Anat Hafte 10:287-305,1898. 382. Nelson DW, Gingrass RP. Anatomy of the mandibular branches of the facial nerve. Plast Reconstr Surg 64:479-482,1979. 383. Nesathurai S, Harvey DT, Schatz SM. Gustatory sweating subsequent to upper thoracic sympathectomy. Arch Phys Med Rehabil 76:104-107,1995. 384. Nesci E, Motta P. Distribution of the extracranial portion of the facial nerve in man. Anat Anz 131:82- 87,1972. 385. New GB, Bozer HE. Hyperhydrosis of the cheek associated with the parotid region. Minn Med 5:652- 657,1922. 386. Nichols RD, Stine PH, Bartschi LR. Facial nerve function in 100 consecutive parotidectomies. Laryngoscope 89:1930-1934,1979. 387. Nicolai A. An early account of gustatory sweating (Frey's syndrome): a chance observation 250 years ago. Br J Plast Surg 38:122-123,1985. 388. Nigro MF, Spiro RH. Deep lobe parotid tumors. Am J Surg 134:523-527,1977. 389. Niparko JK, Beauchamp ML, Krause CJ, Baker SR, Work WP. Surgical treatment of recurrent pleomorphic adenoma of the parotid gland. Arch Otolaryngol Head Neck Surg 112:1180-1184,1986. 390. Norman JEdeB. History of the salivary glands and mixed parotid tumor. In Color atlas and text of the salivary glands. Diseases, disorders and surgery. Eds: Norman JEdeB, McGurk M. London, Mosby- Wolfe. pp:1-12,1995. 391. Nosan KD, Ochi JW, Davidson TM. Preservation of facial contour during parotidectomy. Otolaryngol Head Neck Surg 104:293-298,1991. 392. Novotny GM, Pirozynski WJ. Tumors of the major salivary glands. Review of 100 consecutive cases treated at the Royal Victoria Hospital. Laryngoscope 78:2160-2169,1968. 393. Nukada H, Pollock M, Allpress S. Experimental cold injury to pripheral nerve. Brain 104:779-811,1981. 394. O'Brien CJ, Malka VB, Mijailovic M. Evaluation of 242 consecutive parotidectomies performed for benign and malignant disease. Aust NZ J Surg 63:870-877,1993.

220 395. Ochoa J, Fowler TJ, Gilliatt RW. Anatomical changes in periperal nerves compressed by a pneumatic tourniquet. J Anat 113:433-455,1972. 396. O'Dwyer PJ, Farrar WB, Finkelmeier WR, McCabe DP, James AG. Facial nerve sacrifice and tumor recurrence in primary and recurrent benign parotid tumors. Am J Surg 152:442-445,1986. 397. Ogata K, Naito M. Blood flow of peripheral nerve effects of dissection, stretching and compression. J Head Surg 11:10-14,1986. 398. Ohyama M, Obata E, Furuta S, Sakamoto K, Ohbori Y, Iwabuchi Y. Face EMG topographic analysis of mimetic movements in patients with Bell's palsy. Acta Otolaryngol (Stockh) Suppl. 446:47-56,1988. 399. Olsen KD, Daube JR. Intraoperative monitoring of the facial nerve: an aid in the management of parotid gland recurrent pleomorphic adenomas. Laryngoscope 104:229-232,1994. 400. Osborne DH, Keane FB, Browne HJ, McCollum ST. Superficial parotidectomy for mixed cell tumours of the parotid. Irish Med J 73:420-422,1980. 401. Owen ERTC, Banerjee AK, Kissin M, Kark AE. Complications of parotid surgery: the need for selectivity. Br J Surg 76:1034-1035,1989. 402. Owsley JQ. Platysma-facial rhytidectomy: a preliminary report. Plast Reconstr Surg 60:843-850,1977. 403. Owsley JQ. SMAS-platysma face lift. Plast Reconstr Surg 71:573-576,1983. 404. Ozersky D, Baek SM, Biller HF. Percutaneous identification of the temporal branch of the facial nerve. Aesth Plast Surg 4:276,1980. 405. Palva A, Jokinen K. Tumours of the parotid region. J Laryngol Otol 89:419-425,1975. 406. Parekh D, Glezerson G, Stewart M, Esser J, Lawson HH. Post-traumatic parotid fistulae and sialoceles. Ann Surg 209:105-111,1989. 407. Patey DH. The treatment of mixed tumours of the parotid gland. Brit J Surg 28:29-38,1940. 408. Patey DH. The present position of parotidectomy in the surgery of the parotid gland. Arch Middlesex Hosp 4:91-105,1954. 409. Patey DH. Risk of facial paralysis after parotidectomy. Brit Med J ii:1100-1102,1963. 410. Patey DH. Tumours and other diseases of the salivary glands in relation to general physiology and pathology. J Laryngol Otol 82:853-866,1968. 411. Patey DH, Moffat W. A clinical and experimental study of functional paralysis of the facial nerve following conservative parotidectomy. Brit J Surg 48:435-442,1961. 412. Patey DH, Ranger I. Some points in the surgical anatomy of the parotid gland. Brit J Surg 45:250- 258,1957. 413. Patey DH, Thackray AC. The pathological anatomy and treatment of parotid tumors with retropharyngeal extension (dumb-bell tumors). Brit J Surg 44:352-358,1957. 414. Patey DH, Thackray AC. The treatment of parotid tumours in the light of a pathological study of parotidectomy material. Br J Surg 45:477-487,1958. 415. Payne RT. Pneumococcal parotiditis and antecedent auriculotemporal syndrome. Lancet i:634- 636,1940. 416. Peckitt NS. The facial nerve function coefficient: Analysis of 100 normal subjects. J Oral Maxillofac Surg 50:338-339,1992. 417. Pedersen D, Overgaard J, Sogaard H, Elbrond O, Overgaard M. Malignant parotid tumors in 110 consecutive patients: treatment results and prognosis. Laryngoscope 102:1064-1069,1992. 418. Peitersen E. Natural history of Bell's palsy. Am J Otol 4:107-111,1980. 419. Pfeiffer W, Gellis SS. Auriculotemporal syndrome. Pediatrics 7:670-678,1951. 420. Piorkowski RJ, Guillamondegui OM. Is aggressive surgical treatment indicated for recurrent benign mixed tumors of the parotid gland. Am J Surg 142:434-436,1981.

221 421. Pitanguy I, Ramos AS. The frontal branch of the facial nerve: the importance of its variations in face liftings. Plast Reconstr Surg 38:352-356,1966. 422. Pitois M, Renard JC. Embryologie de la glande parotide. Rev Laryngol (Bordeaux) 100:535-546,1979. 423. Potdar GG, Paymaster JC. Tumors of the salivary glands. Am J Surg 118:440-447,1969. 424. Proctor B. The extratemporal facial nerve. Otolaryngol Head Neck Surg 92:537-545,1984. 425. Quinodoz D, Dulguerov P, Cosenday G, Arechalde A, Piletta P. Traitement du syndrome de Frey avec la toxine botulinique. Méd et Hyg 55:2070-2073,1997. 426. Ramadan HH, Wax MAK, Itani M. The Shaw scalpel and development of facial nerve paresis after superficial parotidectomy. Arch Otolaryngol Head Neck Surg 124:296-298,1998. 427. Randall WC. Sweat gland activity and changing patterns of sweat secretion on the skin surface. Am J Physiol 147:391-398,1946. 428. Rappaport I, Allison GR. Superficial musculoaponeurotic system amelioration of parotidectomy defects. Ann Plast Surg 14:315-323,1985. 429. Rawson AJ, Howard JM, Royster HP, Horn RC. Tumors of the salivary glands: A clinicopathological study of 160 cases. Cancer 3:445-458,1950. 430. Raymond L, Obradovic M, Fioretta G, Droin N, Bouchardy C: Le cancer à Genève. Incidence, mortalité, survie. Genève, Registre genevois des tumeurs.1993. 431. Raymond P. Des éphidroses de la face I-III. Arch Neurol (Paris) 15:51-75,1888. 432. Raymond P. Des éphidroses de la face IV. Arch Neurol (Paris) 15:212-227,1888. 433. Rebierre P, Thomé P. Hydrorhée faciale par excitations diverses et de topographie variable suivant la nature et le siège de l'excitant du sympathique. Rev Oto-Neuro-Ophtalmol 10:347-350,1932. 434. Redleaf MI, McCabe BF. Gustatory otorrhea: Frey's syndrome of the external auditory canal. Ann Otol Rhinol Laryngol 102:438-440,1993. 435. Redon H. Technique de la parotidectomie totale avec conservation du nerf facial. J Chir (Paris) 61:15- 20,1945. 436. Redon H. Discussion on the surgical treatment of parotid tumours. Proc Roy Soc Med 46:1013- 1020,1953. 437. Redon H. Chirurgie des glandes salivaires. Paris, Masson, 1955. 438. Rees TD, Aston SJ, Thorne CHM. Blepharoplasty and facialplasty. In Plastic Surgery. Eds: McCarthy JG. Philadelphia, W.B. Saunders Company. pp:2320-2414,1990. 439. Reissner D. Surgical procedure in tumors of the parotid gland. Arch Surg 65:831-848,1952. 440. Richardson G, Dickason W, Gaisford J, Hanna D. Tumors of salivary gland. Plast Reconstr Surg 55:131- 138,1975. 441. Richins CA, Kuntz A. Role of sympathetic nerves in the regulation of salivary secretions. Brain Res 173:471-473,1953. 442. Richter M, Dulguerov P, Pittet B, Becker M. Immediate function of the temporo-mandibular joint after total resection and reconstruction. J Cranio Maxillofac Surg 8:383-390,1997. 443. Robertson MS, Blake P. A method of using the tympanomastoid fissure to find the facial nerve at parotidectomy. Aust NZ J Surg 54:369-373,1984. 444. Rodel R, Lang J. Untersuchungen zum Verlauf des Ramus marginalis mandibulae nervi fascialis. Laryngo Rhino Otol 75:368-371,1996. 445. Rodriguez-Bigas MA, Sako K, Razack MS, Shedd DP, Bakamijian VY. Benign parotid tumors: a 24-year experience. J Surg Oncol 46:159-161,1991. 446. Ross BG, Fradet G, Nedzelski JM. Development of a sensitive clinical grading system. Otolaryngol Head Neck Surg 114:380-386,1996.

222 447. Ross JAT. The function of the tympanic plexus as related to Frey's syndrome. Laryngoscope 80:1816- 1833,1970. 448. Rousou JA, Parker T, Engelman RM, Breyer RH. Phrenic nerve paresis associated with the use of iced slush and the cooling jacket for topical hypothermia. J Thorac Cardiovasc Surg 89:921-925,1985. 449. Rouyer J. Note sur l'éphidrose parotidienne. Journal de la Physiologie de l'homme et des animaux 2:447-449,1859. 450. Ruaux C, D'Albertas DL, Serrano E, Pessey JJ, Lacomme Y. Atteinte motrice faciale après chirurgie de la parotide. Ann Otolaryngol Chir Cervicofac 111:161-167,1994. 451. Rubin LR. The paralyzed face. St Louis, Mosby - Year Book Inc., 1991. 452. Rudolph R. Depth of the facial nerve in face lift dissections. Plast Reconstr Surg 85:537-544,1990. 453. Sajjadian A, Isaacson G. Electrosurgery in the head and neck. Ann Otol Rhinol Laryngol 107:254- 261,1998. 454. Saksela E, Tarkkanen J, Kohonen A. The malignancy of mixed tumors of the parotid gland. Acta Otolaryngol (Stockh) 70:62-70,1970. 455. Saltzstein HC. Total removal of parotid gland with preservation of facial nerve. Ann Surg 103:635- 638,1936. 456. Samson MJ, Metson R, Wang CC, Montgomery WW. Preservation of the facial nerve in the management of recurrent pleomorphic adenoma. Laryngoscope 101:1060-1062,1991. 457. Sargent EW, Fadhli OA, Cohen RS. Measurement of facial movement with computer software. Arch Otolaryngol Head Neck Surg 124:313-318,1998. 458. Sato K. The physiology, pharmacology and biochemistry of the eccrine sweat gland. Rev Physiol Biochem Pharmacol 79:51-131,1977. 459. Sato K, Dobson RL. Regional and individual variations in the function of the human eccrine sweat gland. J Invest Derm 54:443-449,1970. 460. Sato K, Kang WH, Saga K, Sato KT. Biology of sweat glands and their disorders, I: Normal sweat gland function. J Am Acad Dermatol 20:537-563,1989. 461. Sato K, Kang WH, Saga T, Sato KT. Biology of sweat glands. II: Disorders of sweat gland function. J Am Acad Dermatol 20:699-700,1989. 462. Sato KT, Richardson A, Timm DE, Sato K. One step iodine starch method for direct visualization of sweating. Am J Med Sci 295:528-531,1988. 463. Schaumburg H, Byck R, Herman R, Rosengart C. Peripheral nerve damage by cold. Arch Neurol 16:103- 109,1967. 464. Schaumburg HH, Berger AR. Human toxic neuropathy due to industrial agents. In Peripheral neuropathy. Eds: Dyck PJ, Thomas PK. Philadelphia, WB Saunders. pp:1533-1548,1993. 465. Schliack H, Schiffter R, Goebel HH, Schiffter-Retzlaw I. Untersuchungen zur Frage der Schweissdrüsseninnervation im Bereich des Gesichts. Acta Anat 81:421'438,1972. 466. Schmelzer JD, Zochodne D, Low P. Ischemic and reprerfusion injury of rat peripheral nerve. Proc Natl Acad Sci USA 86:1639-1642,1989. 467. Schnarch A, Markitziu A. Dysguesia, gustatory sweating, and crocodile tears syndrome induced by a cerebellopontine angle meningioma. Oral Surg Oral Med Oral Pathol 70:711-714,1990. 468. Schuknecht HF, Gulya AJ. Anatomy of the temporal bone with surgical implications. Philadelphia, Lea & Febiger, 1986. 469. Schulze-Bonhage A, Schroder M, Ferbert A. Botulinum toxin in the therapy of gustatory sweating. J Neurol 243:143-146,1996. 470. Schwember G, Rodriguez A. Anatomic surgical dissection of the extraparotid facial nerve. Plast Reconstr Surg 81:183-188,1988.

223 471. Scott BA, Sternberg CM. Deep neck space infections. In Head and Neck Surgery - Otolaryngology. Eds: Bailey BJ, Pillsbury HC. Philadelphia, J.B. Lippincott. pp:738-753,1993. 472. Seddon HJ. Surgical disorders of the peripheral nerves. Edinburgh, Churchill Livingstone, 1975. 473. Sessions RB, Roark DT, Alford BR. Frey's syndrome - a technical remedy. Ann Otol 85:734-739,1976. 474. Shaari CM, Sanders I. Assessment of the biological activity of botulinum toxin. In Therapy with botulinum toxin. Eds: Jankovic J, Hallett M. New York, Marcel Drekker Inc. pp:159-170,1994. 475. Shelley WB, Horvath PN. Comparative study on the effect of anticholinergic compounds on sweating. J Invest Dermatol 16:267-274,1951. 476. Shemen LJ. Expanded polytef reconstructing postparotidectomy defects and preventing Frey's syndrome. Arch Otolaryngol Head Neck Surg 121:1307-1309,1995. 477. Shockley WW. Parotidectomy. In The Neck. Diagnosis and surgery. Eds: Shockley WW, Pillsbury HC III. St Louis, Mosby. pp:513-525,1994. 478. Sieffert G, Brocheriou C, Cardesa A, Eveson JW. WHO International histological classification of tumours. Tentative histological classification of salivary gland tumours. Path Res Pract 186:555- 581,1990. 479. Singleton GT, Cassisi NJ. Frey's syndrome: incidence related to skin flap thickness in parotidectomy. Laryngoscope 90:1636-1638,1980. 480. Sinha BK, Buntine DW. Parotid gland tumors. Clinicopathological study. Am J Surg 129:675-679,1975. 481. Sistrunk WE. Mixed tumors of the parotid gland. Minnesota Medicine 4:155-160,1921. 482. Skolnik E, Friedman M, Becker S, Sisson GA, Keyes GR. Tumors of the major salivary glands. Laryngoscope 87:843-861,1977. 483. Skoog T. Plastic Surgery - New methods and refinements. Philadelphia, W.B. Saunders Company, 1974. 484. Sly RM. Auriculotemporal syndrome. Cutis 28:423-425,1981. 485. Smith IM, Murray JAM, Cull RE, Slattery J. A comparison of facial grading systems. Clin Otolaryngol 17:303-307,1991. 486. Smith IM, Murray JAM, Cull RE, Slattery J. Facial weakness. A comparison of clinical and photographic methods of observation. Arch Otolaryngol Head Neck Surg 117:906-909,1991. 487. Spinell C, Ricci E, Berti P, Miccoli P. Postoperative salivary fistula: therapeutic action of octreotide. Surgery 117:117-118,1995. 488. Spiro RH. Salivary neoplasms: overview of a 35 years experience with 2807 patients. Head Neck Surg 8:177-184,1986. 489. Spiro RH, Armstrong J, Harrison L, Geller NL, Lin SY, Strong EW. Carcinoma of major salivary glands. Recent trends. Arch Otolaryngol Head Neck Surg 115:316-321,1989. 490. Spiro RH, Huvos AG, Strong EW. Cancer of the parotid gland: a clinicopathological study of 288 primary cases. Am J Surg 130:452-459,1975. 491. Spiro RH, Martin H. Gustatory sweating following parotid surgery and radical neck dissection. Am Surg 165:118-127,1967. 492. State D. Superficial lobectomy and total parotidectomy with preservation of the facial nerve in the treatment of parotid tumors. Surg Gynecol Obst 237-241,1949. 493. Stea G. Conservative surgical treatment of mixed tumours of the parotid gland. J Max Fac Surg 3:135- 137,1975. 494. Stein I, Geschickter CF. Tumors of the parotid gland. Arch Surg 28:482-526,1934. 495. Stennert E, Vollrath M, Schulte FJ. Parese- und Defectheilings Index. Ein leicht anwendbares Schema zur objectiven Bewertung von Therapieerfolgen bei Fazialisparesen. HNO 24:386-395,1977. 496. Stenonius N. De glandulis oris et novis earundum vasis. Amsterdam, 1661.

224 497. Stevens HE, Doyle PJ. Bilateral gustatory rhinorrhea following bilateral parotidectomy: a case report. J Otolaryngol 17:191-193,1988. 498. Stevens KL, Hobsley M. The treatment of pleomorphic adenomas by formal parotidectomy. Br J Surg 69:1-3,1982. 499. Stewart FW. The diagnosis of tumors by aspiration biopsy. Pathol 9:801-812,1933. 500. Stoll W, Buhner U, T. D. Prinzipien der Parotischirurgie. Retrospektive Studie an 1009 Parotidektomien. Chir Praxis 34:487-495,1985. 501. Stuart DD. Diabetic gustatory sweating. Annals of Internal Medicine 89:223-224,1978. 502. Stuzin JM, Baker TJ, Gordon HL. The relationship of the superficial and deep facial fascias: relevance to rhytidectomy and aging. Plast Reconstr Surg 89:441-449,1992. 503. Stuzin JM, Wagstrom L, Kawamoto HK, Wolfe SA. Anatomy of the frontal branch of the facial nerve: the significance of the temporal fat pad. Plast Reconstr Surg 83:265-271,1989. 504. Sultan MR, Wider TM, Hugo NE. Frey's syndrome: prevention with temporoparietal fascia. Ann Plast Surg 34:292-296,1995. 505. Sunderland S. Nerves and nerve injuries. Edinburgh, Chirchill Livingstone, 1978. 506. Sweat M. Cases of extirpation of parotid glands. NY J Med 7:23-26,1851. 507. Swinton NW, Warren S. Salivary gland tumors. Surg Gynecol Obst 67:424-435,1938. 508. Tabb HG, Scalco AN, Floyd FS. Exposure of the facial nerve in parotid surgery. Laryngoscope 80:559- 567,1970. 509. Tainino H, Vaalasti A, Rechardt L. The distribution of substance P-, CGRP-, galanin-, and ANP-like immunoreactive nerves in human sweat glands. Histochem J 19:375-380,1985. 510. Takahashi H, Izumi H, Karita K. Parasympathetic reflex salivary secretion in the cat parotid gland. Jap J Physiol 45:475-490,1995. 511. Terrell JE, Kileny PR, Yian C, Escalamado RM, Bradford CR, Pillsbury MS, Wolf GT. Clinical outcome of continuous facial nerve monitoring during parotidectomy. Arch Otolaryngol Head Neck Surg 157:1081- 1087,1997. 512. Terris DJ, Tuffo KM, Fee WE. Modified facelift incision for parotidectomy. J Laryngol Otol 108:574- 578,1994. 513. Tessier P. Le SMAS, sa petite histoire. Ann Chir Plast 26:191-192,1981. 514. Testut L. Traité d'anatomie humaine. Paris, Octave-Doin, 1901. 515. Thackray AC, Sobin LH. Histological typing of salivary gland tumors. Geneva, World Health Organization, 1972. 516. Thaller SR, Kim S, Patterson H, Widman M, Daniller A. The submuscular aponeurotic system (SMAS): a histologic and comparative anatomy. Plast Reconstr Surg 86:690-696,1990. 517. Thomas A. Le double réflexe vaso-dilatateur et sudoral de la face consécutif aux blessures de la loge parotidienne. Rev Neurol (Paris) 1:447-460,1927. 518. Tortella EP. Le plexus parotidien du facial. Ann Anat Pathol (Paris) 12:41-50,1935. 519. Treves F. Surgical applied anatomy. London, Cassell, 1907. 520. Trotoux J. Parotidectomies. Paris, Arnette, 1992. 521. Trotoux J. Technique de comblement per-opératoire de la dépression rétro-mandibulaire après parotidectomie: Utilisation d'un caillot sanguin. Ann Otolaryngol Chir Cervicofac 111:35-37,1994. 522. Trotoux J, Vilde F, Kossowski M, Clavier A, Khayat M, Londero A, Bonfils P. Reprises chirurgicales des tumeurs de la parotide, approche technique et consequences pour le VII. J Chir (Paris) 130:349- 357,1993.

225 523. Tuinzing DB, Van der Kwast WAM. Frey's syndrome; a complication after sagittal splitting of the mandibular ramus. Int J Oral Surg 11:197-200,1982. 524. Turner JC, Devine KD, Judd ES. The auriculotemporal (Frey) syndrome occurring after parotid surgery. Surg Gynecol Obst 564-568,1960. 525. Uno H. Sympathetic innervation of the sweat glands and piloerector muscle of macaques and human beings. J Invest Dermatol 69:112-130,1977. 526. Urken ML, Biller HF. Sternocleidomastoid. In Atlas of regional and free flaps for Head and Neck reconstruction. Eds: Urken ML, Cheney ML, Sullivan MJ, Biller HF. New York, Raven Press. pp:49- 64,1995. 527. Utendorfer RW. Total parotidectomy for mixed tumors. Am J Surg 91:159-163,1956. 528. Van Miert PJ, Dawes JDK, Harkness DG. The treatment of mixed parotid tumors: A report of 183 cases. J Laryngol Otol 82:459-468,1968. 529. Van Niiekerk JLM, Wobbes T, Monstrey S, Bruaset I. The management of parotid tumors; a ten years experience. Acta Chir Belg 87:1-5,1987. 530. Vandenberg HJ, Kambouris A, Pryzybylski T, Rachmaninoff N. Salivary tumors. Clinicopathological review of 190 patients. Am J Surg 108:480-484,1964. 531. VanSwearingen JM, Brach JS. The facial disability index: reliability and validity of a disability assessment instrument for disorders of the facial neuromuscular system. Phys Ther 76:1288-1300,1996. 532. Vilain R, Mity V. Les nerfs de l'oreille et la perte de sensibilité du lobule de l'oreille dans les suites des incisions de la région parotidomastoidienne. Ann Chir Plast 21:75-82,1976. 533. Wallis KA, Gibson T. Gustatory sweating following parotidectomy: correction by a fascia lata graft. Brit J Plast Surg 31:68-71,1978. 534. Wang D, Perneger TV, Lehmann W, Dulguerov P. Mimique normale et pathologique : analyse objective. Scheiz Med Wochenschrift In press. 535. Wang DZ, Liv SJ, Donoff RB, Guralnick W. A modified centripetal approach to parotidectomy. J Oral Maxillofac Surg 43:14-19,1985. 536. Ward CM. Injury of the facial nerve during surgery of the parotid gland. Br J Surg 62:401-403,1975. 537. Wassef M. Superficial fascial and muscular layers in the face and neck: a histological study. Aesth Plast Surg 11:171-176,1987. 538. Watanabe Y, Ishikawa M, Shojaku H, Mizukoshi K. Facial nerve palsy as a complication of parotid gland surgery and its prevention. Acta Otolaryngol 504:137-139,1993. 539. Watkin GT, Hobsley M. Influence of local surgery and radiotherapy on the natural history of pleomorphic adenomas. Br J Surg 73:74-76,1986. 540. Watkins J. Facial sweating after food: a new sign of diabetic autonomic neuropathy. Br Med J i:583- 587,1973. 541. Watson C, Vijayan N. The sympathetic innervation of the eyes and the face. Clin Anat 8:262-272,1995. 542. Wax M, Tarshis L. Post-parotidectomy fistula. J Otolaryngol 20:10-13,1991. 543. Weber FP. Clinical cases V: a case of localized sweating and blushing on eating, possibly due to temporary compression of vasomotor fibers. Trans Clin Soc London 31:277-280,1897. 544. Welti H. A propos de 35 tumeurs parotidiennes opérées. Mém Acad Chir 81:1007-1011,1955. 545. Wennmo C, Spandow O, Emgard P, Krouthen B. Pleomorphic adenoma of the parotid gland: superficial parotidectomy or limited excision? J Laryngol Otol 102:603-605,1988. 546. Wharton T. Adenographia. Siva glandularum totius corporis descriptio. Amsterdam, Sumptibus Joannis Ravesreinii, 1659. 547. Wiegand DA, Fickel V. Acoustic neuroma - the patient's perspective: subjective assessment of symptoms,diagnosis, therapy, and outcome in 541 patients. Laryngoscope 99:179-187,1989.

226 548. Wilbourn AJ. Iatrogenic nerve injuries. Neurol Clinics North Am 16:55-82,1998. 549. Wilson WC. Observations relating to the innervation of the sweat glands of the face. Clin Sci 2:273- 286,1936. 550. Wilson-Pauwels LW, Akesson EJ, Stewart PA. Cranial nerves. Anatomy and clinical comments. Philadelphia, B. C. Decker, 1988. 551. Windebank AJ. Metal neuropathy. In Peripheral neuropathy. Eds: Dyck PJ, Thomas PK. Philadelphia, WB Saunders. pp:1549-1570,1993. 552. Winsten J, Ward GE. The parotid gland: an antomical study. Surgery 40:585-606,1956. 553. Winsten J, Ward GE. Mixed tumors of the parotid gland. Surgery 42:1029-1035,1957. 554. Wolf SR, Schneider B, Suchy B, Eichhorn B. Intraoperatives Fazialisnervmonitoring in der Parotischirurgie. HNO 43:294-298,1995. 555. Wood DA, Hughes GB, Sesic M, Good TL. Objective measurement of normal facial movement with video microscaling. Am J Otol 15:61-65,1994. 556. Woods JE. Parotidectomy: points of technique for brief and safe operation. Am J Surg 145:678- 683,1983. 557. Woods JE. Parotidectomy versus limited resection for benign parotid masses. Am J Surg 149:749- 750,1985. 558. Woods JE, Chong GC, Beahrs OH. Experience with 1360 primary parotid tumors. Am J Surg 130:460- 462,1975. 559. Work WP, Batsakis JG, Bailey DG. Recurrent benign mixed tumor and the facial nerve. Arch Otolaryngol 102:15-19,1976. 560. Yamashita T, Tomoda K, Kumazawa T. The usefulness of partial parotidectomy for benign parotid gland tumors: A retrospective study of 306 cases. Acta Otolaryngol (Stockh) Suppl 500:113-116,1993. 561. Yanagihara N. Grading facial palsy. In Facial nerve surgery. Proceedings: Third international symposium on facial nerve surgery. Eds: Fisch U. Amstelveen, Kugler Medical Publications. pp:553- 535,1977. 562. Yu LT, Hamilton R. Frey's syndrome: prevention with conservative parotidectomy and superficial musculoaponeurotic system preservation. Ann Plast Surg 29:217-222,1992. 563. Yuen K, Inokuchi I, Maeta M, Kawakami SI, Masuda Y. Evaluation of facial palsy by moiré topography index. Otolaryngol Head Neck Surg 117:567-572,1997. 564. Zahejsky J, Rovensky J. A comparison of the effectiveness of several external antiperspirants. J Soc Cosmetic Chem 23:775-789,1972. 565. Ziarah HA, Atkinson ME. The angular tract : an important anatomical structure of surgical significance. Br J Oral Surg 19:116-123,1981. 566. Ziarah HA, Atkinson ME. The surgical anatomy of the mandibular distribution of the facial nerve. Br J Oral Surg 19:159-170,1981. 567. Zigliotti GL, Liverani MB, Ghibellini D. The relationship between parotid and superficial fasciae. Surg Radiol Anat 293-300,1991. 568. Zoa Chang E, Lee WC. Surgical treatment of pleomorphic adenoma of the parotid gland: report of 110 cases. J Oral Maxillofac Surg 43:680-682,1985. 569. Zoller J, Herrmann A, Maier H. Frey's syndrome secondary to a subcondylar fracture. Otolaryngol Head Neck Surg 108:751-753,1993.

227