The Spread and Staging of Cervical Cancer

Although surveillance with cervical cytologic smears has resulted in a significant decrease in deaths from invasive cervical cancer in industrialized nations, a significant number of women continue to develop advanced carcinoma of the cervix each year in the United States. In 2001, an estimated 12,900 cases of invasive cervical cancer will have occurred in the country, and more than 4,400 women will have died from cervical cancer.¹ The epidemiology of this disease was reviewed in an earlier chapter. However, at the outset, it should be emphasized that a subgroup of patients with invasive cervical cancer are at greater risk for a more aggressive cervical cancer and for more advanced stage. These patients include those with immunocompromised states, such as renal transplant patients or those who test positive for HIV on serum testing. Although workup and studies are indicated for all patients with clinically advanced disease, immunocompromised patients should have extensive workups before treatment modalities are chosen.

The purpose of this chapter is to review the usual patterns of spread of cervical cancer. Traditional and more modern pretreatment diagnostic modalities will be explored.

Direct local extension and lymphatic embolization are the primary routes of spread of cervical carcinoma. Hematogenous dissemination usually occurs with more advanced disease or unusual cell types, such as adenosquamous or neuroendocrine tumors. The progression of squamous cell neoplasias to invasive carcinomas was addressed in earlier chapters. If the basement membrane is intact and stromal invasion is absent, dissemination and metastases do not occur.

Direct Extension

Cervical cancer usually extends by lateral spread to involve the parametria, following a path of least resistance. In this scenario, the cardinal ligaments eventually are involved. Another common pattern of spread is distal, to involve the vaginal fornices and later (and with larger-volume tumors) the midportion and distal third of the vagina. It may spread posteriorly to involve the rectum or the uterosacral ligaments. Rectal spread is usually associated with posterior vaginal involvement. Anterior spread to the bladder is unusual in the absence of large-volume tumors with parametrial extension.

These observations are important in the decision to order expensive tests in this era of cost containment. As will later be enumerated, most traditional testing can be avoided in early-stage diseases, particularly when operative intervention is planned.

Lateral spread to the parametria mandates some type of radiologic evaluation of the ureters. The anatomic position of the ureter (within 2 cm of the uterine cervix) makes the ureter vulnerable to involvement. About one third of patients with stage IIIB cancers have ureteral obstruction, and about 5% have bilateral obstruction.² The obstruction usually occurs in the distal third of the ureter.

Lymphatic Embolization

Lymphatic spaces are present within 1 to 2 mm of the basement membrane. If these spaces are involved, lymphatic embolization to regional lymph nodes can occur. It is more likely to occur with large-volume tumors and if several lymphatic spaces are involved. The lymphatic channels draining the cervix converge to form lateral trunks. These course through the parametria and drain to the external iliac, hypogastric, obturator, and common iliac nodes (Fig. 1).³ Also, small anterior channels pass behind the bladder and terminate in the external iliac nodes. Posterior channels drain directly into the common iliac and para-aortic nodes and superior rectal nodes.

Para-aortic nodal spread is usually by an orderly, progressive involvement of the pelvic and common iliac nodes. However, cervical cancer can occasionally metastasize directly to the para-aortic nodes by embolization through the posterior trunks. As noted in Figure 1, three para-aortic node chains exist. About 6 to 10 aortic nodes are located to the left, anterior and posterior to the aorta. The aorta caval nodes (2 to 5) lie between the aorta and the vena cava, and the caval nodes (10 to 15) are located to the right and anterior to the vena cava.^3,4 Lymph is transported cephalad to the cisterna chyli to the thoracic duct and eventually to the scalene nodes (Fig. 2). It also empties into the venous system. Careful examination of the supraclavicular areas is mandatory when evaluating patients with cervical cancer.

Hematogenous Spread

Veins and lymphatic spaces lie close to the basement membrane. Patients with large lesions and more advanced disease may have hematogenous dissemination by direct blood vessel invasion, by way of lacerated capillaries and veins, through the thoracic or through smaller lymphatic and venous channels. Blood vessel invasion usually occurs in veins rather than arteries.⁵ About 5% of patients with cervical cancer have hematogenous spread.

Poorly differentiated tumors and aggressive cell types may be more likely to spread by the hematogenous route. In one reported series of patients with adenosquamous cell carcinoma, 88% had distant metastases at the time of death.⁶ Survival in this retrospective study, independent of lesion size, was only 27% for patients with stage IB lesions. Others have not noted this trend for poor survival with this cell type, nor the tendency to have distant blood-borne metastases.^7,8 However, in a recent analysis of 813 patients with stage IB disease managed by radical surgery, Look and associates⁹ noted a poorer survival for patients with adenosquamous lesions. Also, more patients developed extrapelvic recurrences than those with squamous or adenocarcinoma cell types. Neuroendocrine cervical tumors and glassy cell tumors have also been associated with hematogenous spread with early-stage disease.

About 1% to 2% of patients with cervical carcinomas present with lung metastases, and 5% to 35% eventually develop pulmonary metastases.^10,11 Other common sites of blood-borne metastases include liver (3%),³ bone (16%),¹¹ and bowel. Bone metastases can be due to local extension, although distant metastases are usually associated with hematogenous dissemination. These lesions are usually radiographically lytic, and patients usually have recurrent or advanced disease or other sites of metastases.¹² Small bowel metastases can be the result of direct extension from para-aortic node involvement or from intraperitoneal dissemination.³

Awareness of certain high-risk cell types for relatively early distant metastases can help in more appropriate ordering of tests in this cost-conscious era. Dissemination generally follows an orderly sequence. Finally, direct extension to the lower uterine segment and/or endometrial cavity occurs in about 10% to 30% of patients.¹³

Most gynecologic and radiation oncologists in the United States report their results using the International Federation of Gynecology and Obstetrics (FIGO) staging system. The initial staging system was proposed by a subcommittee of the League of Nations in 1929 and was revised in 1937 and 1950. FIGO accepted the League of Nations staging in 1950. The current clinical staging system of FIGO has changed minimally since 1962. Ureteral obstruction was found to have an adverse effect on survival, and this was reflected in the 1971 FIGO modifications.¹⁴ The FIGO classification was again changed in 1974: with a better understanding of the effect of an occult lesion on prognosis, frankly invasive cancers were removed from stage IA. Stage IV was also subdivided into cases with bladder or rectal invasion. Stage IVB was assigned to patients with distant metastases.

In 1974, the Society of Gynecologic Oncologists (SGO) proposed the definition of microinvasion as stromal invasion 3 mm or less below the basement membrane and absence of capillary lymphatic (C/L) space involvement.¹⁵ As will be noted, microinvasive carcinoma of the cervix has been poorly defined in the past and is still a focus of persistent controversy.

Most of the changes in FIGO staging in the past decade have occurred in stage I disease. In 1985, FIGO defined stage IA as “preclinical invasive carcinoma, diagnosed by microscopy only.” Stage IA was subdivided into IA1 (minimal microscopic stromal invasion) and IA2 (tumor with invasive component 5 mm or less taken from the base of the epithelium and 7 mm or less horizontal spread).¹⁶ C/L space involvement did not exclude patients from being placed in stage IA.¹⁶

In 1995, FIGO further reclassified stage I cervical cancer. Stage IA was subdivided by depth of stromal invasion in an attempt to delineate the different clinical behaviors and treatments for carcinoma with invasion of less than 3 mm and less than 5 mm. C/L space involvement does not alter the stage but should be recorded. Stage IB was subdivided into lesions less than or greater than 4 cm.^17,18Figure 3 shows a bulky barrel 6-cm cervical carcinoma currently staged as IB2. These carcinomas are sometimes treated with external beam irradiation, one 72-hour cesium, and extrafascial hysterectomy (Fig. 4). The complete 1995 FIGO staging is shown in Table 1. As in the 1985 FIGO staging system, extension to the corpus is disregarded.

Table 1. 1995 FIGO Staging for Carcinoma of the Cervix

FIGO, International Federation of Gynecology and Obstetrics.
Stage IA carcinoma should include minimal microscopically evident stomal invasion as well as small cancerous tumors of measurable size. Stage IA should be divided into those lesions with minute foci of invasion visible only microscopically as stage IA1 and macroscopically measurable microcarcinoma as stage IA2 to gain further knowledge of the clinical behavior of these lesions. The term “IB occult” should be omitted.
The diagnosis of both stages IA1 and IA2 cases should be based on microscopic examination of removed tissue, preferably a cone, which must include the entire lesion. The lower limit of stage IA2 should be measurable macroscopically (even if dots need to be placed on the slide prior to measurement), and the upper limit of stage IA2 is given by measurement of the two largest dimensions in any given section. The depth of invasion should not be more than 5 mm taken from the base of the epithelium, either surface or glandular, from which it originates. The second dimension, the horizontal spread, must not exceed 7 mm. Vascular space involvement, either venous or lymphatic, should not alter the staging but should be specifically recorded, as it may affect treatment decisions in the future.
Lesions of greater size should be classified as stage IB.
A patient with a growth fixed to the pelvic wall by a short and indurated but not nodular parametrium should be assigned to stage IIB. It is impossible, at clinical examination, to decide whether a smooth and indurated parametrium is truly cancerous or only inflammatory. Therefore, the case should be placed in stage III only if the parametrium is nodular on the pelvic wall or if the growth itself extends to the pelvic wall.
The presence of hydronephrosis or nonfunctioning kidney due to stenosis of the ureter by cancer permits a case to be assigned to stage III even if, according to the other findings, the case should be assigned to stage I or II.
The presence of bullous edema, as such, should not permit a case to be allocated to stage IV. Ridges and furrows in the bladder wall should be interpreted as signs of submucous involvement of the bladder if they remain fixed to the growth during palpation (i.e., examination from the vagina or the rectum during cystoscopy). A finding of malignant cells in cytologic washings from the urinary bladder requires further examination and biopsy from the wall of the bladder.

A parallel TNM staging system has been proposed by the American Joint Committee on Cancer (AJCC).¹⁹ The AJCC criteria for the various stages are defined and compared in Tables 2 and 3. All histologic types are included. Unfortunately, the TNM classification is impractical for staging cervical carcinoma because lymphatic or intra-abdominal spread cannot be reliably evaluated clinically. Cervical cancer is one of the few female genital cancers that is not surgically staged.

Table 2. Comparison of Cervical Cancer Staging

AJCC, American Joint Committee on Cancer; FIGO, International Federation of Gynecology and Obstetrics.
*Presence of bullous edema is not sufficient evidence to classify a tumor T4.

Table 3. AJCC Stage Grouping for Carcinoma of the Cervix

AJCC, American Joint Committee on Cancer.

Microinvasion

The controversy regarding staging for squamous cell microinvasion has persisted since the earlier definitions, which were primarily of European origin.^20,21,22,23 Microinvasive carcinoma cannot be defined by biopsy. A conization (or hysterectomy) is needed with 12 or more sections of the cone to rule out foci of deeper invasion. Microinvasive adenocarcinoma is a clinical entity that is not well defined. No data are available to suggest the efficacy of less-than-radical treatment for this entity. In 1983, van Nagell and colleagues ²⁴ described a group of patients who had radical hysterectomy and pelvic lymph node dissections and added their series to the then-available data from the literature. With stromal invasion of less than 3 mm, only 1 of 397 patients had positive pelvic nodes; 8 of 98 patients with invasion of 3.1 to 5 mm had positive nodes. Since that time, several studies have questioned the 1985 FIGO staging because of the relatively high incidence of lymph node metastases in the group of patients with invasion of 3.1 to 5 mm.^25,26,27 Burghardt and associates²⁸ evaluated 486 patients with stage IA lesions, subclassified into 1985 FIGO IA1 and IA2. They concluded that although IA1 lesions could be treated conservatively, the treatment in stage IA2 lesions must be individualized. These worldwide observations, prompted by the SGO definition, led in part to the 1995 FIGO revisions for stage IA.

The issue of C/L space involvement has yet to be resolved for FIGO stage IA. Earlier studies, such as that of Roche and Norris,²⁹ suggested that C/L involvement had little impact on node metastases when patients had microinvasion. The group from M. D. Anderson, however, noted in their review that of patients with less than 3 mm of invasion and no C/L space involvement, 0.3% had pelvic node metastases; 2.6% of those with C/L space involvement had pelvic node metastases.³⁰ Lymph vascular space involvement appears to be a predictor of lymph node metastases. Most gynecologic oncologists in this country continue to use the 1974 SGO definition and will treat only patients with less than 3 mm of invasion and no C/L invasion by conservative surgery.^27,30

The evaluation of treatments based on the 1995 FIGO definition, with C/L space taken into consideration, will be determined by future studies. The importance of C/L space involvement and the question of whether these data should be incorporated into the staging system of stage IA needs to be settled.³¹ However, many Europeans see little need for even the 1995 changes, and as recently noted by Burghardt and colleagues,³² the FIGO classification was never meant to be a treatment guideline.

Clinical Staging

Cervical carcinoma is predominantly a Third World problem. Thus, only certain diagnostic studies are allowed by FIGO, and these tests are available in most countries. A few caveats about clinical staging should be remembered:

1. A biopsy, not a cervical cytologic smear, is necessary to establish the diagnosis.
   
2. The physical examination should include a survey of the skin, careful palpation of lymph node-bearing areas, speculum examination, and a bimanual rectovaginal examination.
   
3. Only the procedures and studies allowed by FIGO can be used in clinical staging.
   
4. Once the stage is determined, it cannot be changed. For instance, a woman who has clinical stage IB and has a metastatic para-aortic node detected at the time of radical hysterectomy still has stage IB disease.
   
5. Patients seen after treatment initiation should be listed as having unstaged cervical carcinoma.
   
6. If cancer remains after therapy has been completed or if invasive cancer is documented within 6 months of treatment conclusion, the patient still has the original stage disease, but it is classified as “persistent” in most institutions.
   

Clinical staging is often inaccurate: 17% to 24% of patients with clinical stage I and 30% to 50% of those with presumed stage II have more advanced disease when surgical staging is done.^33,34 Examination under anesthesia (EUA), particularly when performed by a gynecologic oncologist and a radiation oncologist at the same time, may increase the accuracy of clinical staging by 25%.³⁵ EUA allows a more thorough visual inspection of the upper vagina and a better bimanual examination of the parametria. Patients who have a large-volume tumor and require cystoscopy or proctoscopy may benefit from simultaneous EUA. EUA for staging is used in about 46% of patients who have cervical cancer in the United States,³⁶ but because the procedure is relatively expensive, and because of managed care and cost containment, this figure will no doubt decrease in the future.

FIGO allows the following surgical procedures for staging: cervical biopsy or conization, endocervical curettage, EUA, cystoscopy, and proctoscopy. Biopsy-proven bladder or rectal involvement is uncommon in early-stage, small-volume tumors. However, if the tumor is positioned anteriorly or posteriorly, one of these two studies may be indicated. If proctoscopy or cystoscopy (bullous edema) is abnormal, a biopsy must be done and be positive to place the patient in a stage IVA category. The Patterns of Care Study clearly noted a decrease in the use of cystoscopy (from 38% to 25%) and proctoscopy (from 30% to 19%) in patients with stage IB, IIA disease.³⁶ This retrospective study did not evaluate tumor volume as a criterion for performing or not performing these endoscopic procedures.

Pretreatment blood or serum studies may lead the physician to perform more sophisticated radiographic studies to exclude more advanced disease. An elevated creatinine level suggests urinary tract involvement. Elevated liver function tests suggest liver involvement, and an elevated serum calcium level suggests bone involvement.

Radiographic Studies, Routine

FIGO allows a chest x-ray, intravenous pyelogram (IVP), barium enema, and skeletal x-ray (not a bone scan) to be used for staging. Although metastatic spread to the lung is rare in early-stage disease, it is present in up to 5% of patients with more advanced disease.¹⁰ A chest x-ray should be obtained in all patients. If a solitary nodule is found, histologic evaluation by needle aspiration, or preferably resection by video-assisted thoracotomy or open biopsy (especially in tobacco users), is needed. We have noted a few cervical cancer patients with a second small solitary primary lung cancer who benefited from surgical removal of the lung lesion. IVP studies are less expensive than computed tomography (CT) scans and are abnormal in about 15% of patients.³⁷ Almost one third of patients have an abnormal IVP with stage III disease. IVP studies are not routinely indicated for women with small (less than 2 cm) cervical lesions who are scheduled for radical hysterectomy. An upper gastrointestinal (UGI) series should be ordered only for symptomatic patients. The Patterns of Care Study noted that only 21% of patients with stage IIB through IVB had a barium enema done.³⁶ A barium enema should be ordered only in older women to help identify other colon disease, such as diverticulosis or colon cancer. Radiologic skeletal surveys are rarely ordered, and isotope bone scans should be used for symptomatic patients. Most of the lesions in bone are lytic and can be detected by chest x-ray and IVP.

Radiographic Studies, Advanced

Although the information from these procedures cannot be used to change the FIGO stage, they are very useful in treatment planning. As noted by Heller and associates,³⁸ the lymphangiogram may be a more sensitive test to establish para-aortic node metastases than the ultrasound or CT scan, particularly in women who are at low risk for metastases to these nodes. The authors in this Gynecologic Oncology Group (GOG) study suggested that a negative lymphangiogram may be adequate to eliminate surgical staging in a select group of patients with small-volume tumors. In this prospective study, all 320 patients with negative studies had para-aortic node dissections. Although the lymphangiogram may be more useful than CT in detecting disease in nodes that are smaller, it is often painful and difficult to perform in morbidly obese patients or those with pedal edema.

CT scans are inaccurate for determining parametrial invasion. However, if the nodes are larger than 1.5 cm, para-aortic node metastases can be detected in as many as 91% of women.³⁹ CT scans should be done with intravenous contrast to give optimal information (Fig. 5). However, CT scanning is seldom beneficial in evaluating patients with small-volume disease.⁴⁰ The specificity for detecting lymph node metastases is high, but the sensitivity is low.^38,41 CT is correct in evaluating 92% of lesions staged IIIB through IVB.^42,43,44 Evaluation of enlarged lymph nodes should be complemented by CT-guided fine-needle aspiration. The results of these tests cannot be incorporated into FIGO staging.

The GOG study showed that ultrasound had a sensitivity of only 19% for detecting metastases to para-aortic nodes.³⁸ This indicates that the more expensive CT or magnetic resonance imaging (MRI) studies may have more utility. Ultrasound can be used safely in pregnancy and can be used to evaluate the upper urinary tract. Transrectal ultrasound has been used to evaluate parametrial extension. The sensitivity of transrectal ultrasound in this evaluation was only 78%, with a specificity of 89% and diagnostic accuracy of 87%.⁴⁵ One group has reported that transvaginal color Doppler ultrasound might aid in the diagnosis of early-stage disease.⁴⁶ This technique is expensive. Endosonography produces superior results compared to CT in assessing local extension.⁴⁷ These authors performed the procedure, which consisted of vesical, vaginal, and rectal endosonography, under anesthesia.

At our institution, the cost of imaging a cervical cancer patient with abdominopelvic MRI is about $1,225 versus $1925 for a similar study by CT. MRI should be considered in the evaluation of pregnant patients with invasive cervical cancer. In the past few years, studies in the literature have suggested that MRI is superior to CT in delineating the primary tumor site, tumor dimensions, and the local extent of disease.^48,49,50 Static images enhanced with gadolinium—diethylenetriamine—penta-acetic acid (Gd-DIPA) may not improve staging accuracy but may be useful in selected patients to confirm tumor necrosis or stage IVA disease.^51,52 MRI may not be as useful as CT in detecting nodal metastases; however, in centers with expertise, it may offer more useful information than conventional CT or ultrasound. In a recent study, pretreatment MRI in patients with cervical tumors larger than 2 cm decreased the number of other diagnostic tests ordered, the number of invasive procedures, and the number of unnecessary surgeries; thus, it was considered cost-effective in this setting.⁵³ Routine use of contrast-enhanced MRI is still controversial. MRI, like CT, cannot be used for FIGO staging.

Positron emission tomography (PET), a new technique, is based on the decay of radionuclides and the emission of positively charged particles (positrons). These positrons penetrate short distances and combine with an electron, converting mass into energy. The emission of gamma rays is scanned to reconstruct a three-dimensional representation of tissue events.⁵⁴ Different radionuclides can be used to evaluate tumor metabolism and organ function. PET needs further evaluation, but it has the potential for delineating primary disease more accurately. It may more precisely localize nodal disease.

Fine-needle aspiration should be used to confirm the suspect imaging studies. If the results are positive, radiation treatment fields may merit change (i.e., extended fields for positive para-aortic nodes).⁵⁵ If results are negative, a decision must be made for surgical staging. Fine-needle aspiration is of particular use in aspirating suspect lung lesions, liver lesions, or suspected bone metastases. CT- or MRI-directed aspiration of nodal masses that yields positive results does not change FIGO staging.

Current Serum Studies, Advanced

The so-called serum tumor markers include squamous cell carcinoma antigen (SCCA), carcinoembryonic antigen (CEA), and CA-125. CEA levels have not been found to correlate with clinical stage and are seldom used as a marker for cervical cancer.⁵⁶ CA-125 levels are elevated in some patients with advanced or recurrent cervical adenocarcinoma.^57,58 However, neither CA-125 nor CEA measurement is used in the routine management of cervical cancer patients.

The SCCA level has been found to be elevated in patients with large-volume tumors, deep stromal invasion, and nodal metastases and correlates with treatment response.^57,59 Also, one recent study suggests that the combination of SCCA and CEA measurement may help identify patients needing neoadjuvant chemotherapy in advanced squamous cell carcinoma.⁶⁰ Another recent study suggests that patients with elevated CA-125 and SCCA levels may have more blood vessel involvement in stage IB, IIA disease, and that such patients should be considered for treatments other than radical surgery.⁶¹ Bolger and colleagues⁶² noted a higher incidence of positive nodes in 220 patients treated surgically when the SCCA level was elevated. Elevated SCCA level had no independent prognostic significance.

The National Institutes of Health and the National Cancer Institute held a Consensus Development Conference in Cervical Cancer in April 1996. One of their conclusions was that the “optimal role for imaging studies in defining the extent of disease and in planning radiation therapy needs further investigation, as does the measurement of serum tumor markers in patients with invasive cervical cancer.”⁶³

Surgical Staging: Celiotomy

Clinical staging for cervical cancer is inaccurate. Investigators in the 1970s and 1980s began using surgical staging primarily to assess the status of para-aortic nodes. Some patients with occult para-aortic node metastases are potentially curable when extended-field irradiation is added. The earlier studies of para-aortic node sampling, usually to the level of the inferior mesenteric artery, noted a relatively high incidence of node positivity, increasing as the clinical stage was more advanced (Table 4).^{64,65,66,67,68,69,70} Thus, of patients with stage III disease who received only pelvic irradiation, about one third would receive inadequate irradiation. This progressively higher incidence of positive para-aortic nodes prompted such organizations as the GOG to require para-aortic node sampling before placing patients with advanced cervical cancer on phase 3 studies if imaging modalities did not suggest “suspicious nodes.”

Table 4. Incidence of Para-aortic Node Metastases by Stage in Patients with Cervical Carcinoma, 1977-1981

Also, patients with large, bulky stage IB lesions are treated in some institutions with only pelvic field irradiation. Two studies have noted positive para-aortic nodes in these large stage IB lesions in 20%⁶⁴ and 35%⁷¹ of patients. Podczaski and associates⁷² noted a 14% incidence of positive para-aortic nodes in stage IB cervical cancer patients, but 21 of 35 total patients had bulky cervical disease.

Serious bowel complications have been noted in cervical cancer patients who have had operative evaluation through a transperitoneal approach followed by radiation therapy. In a series of patients who had transperitoneal evaluation of para-aortic nodes, Piver and Barlow⁷³ reported that 21.7% died of radiation-induced bowel injuries. In a similar group of patients who had transperitoneal para-aortic node sampling followed by para-aortic node irradiation, Wharton and colleagues⁷⁰ noted that in 27.6% of the patients, severe intestinal complications occurred later; half the patients with intestinal complications died. In a retrospective study, Berman and associates⁷⁴ noted that of 33 patients operated on by a transperitoneal approach, 30% required operative intervention for later small bowel obstruction. Weiser and colleagues⁷⁵ prospectively analyzed two surgical approaches to para-aortic nodes: transperitoneal and extraperitoneal. No significant differences were found between the two techniques in the sensitivity to detect nodal metastases or the incidence of surgical complications. If the patients had later irradiation, however, both bowel obstruction and other regional enteric injuries were significantly more common in the transperitoneal group. To avoid these complications, an extraperitoneal approach has been advocated.

EXTRAPERITONEAL APPROACH.

Schellhas,⁷⁶ using an upper midline incision, was one of the first to report an extraperitoneal approach to para-aortic nodes. He noted few complications using this small incision (7 to 8 cm), and pelvic portal irradiation could be used without delay. One disadvantage is that it is difficult using this incision to remove the left para-aortic nodes, particularly in obese patients. We previously used and reported on an extraperitoneal midline incision to remove pelvic and para-aortic nodes.⁷⁷ Although no significant complications were noted in this small series, and easy access to both sides of the aorta was obtained, we have abandoned this as a staging procedure because of the delay in initiating radiation therapy.

A modification of the extraperitoneal inguinal incision was described in the gynecologic literature by Berman and associates.⁷⁴ This is actually a modified Gibson incision. The group from UCLA prefers to make the skin incision on the left because the left para-aortic lymphatic channels are lateral and posterior to the aorta. They argue that it is technically more feasible to dissect the precaval nodes from a left-sided incision if a single incision is to be used for extraperitoneal para-aortic node sampling. We prefer to make the incision on the right.

To gain access to the node-bearing areas, an incision is initiated just cephalad to the umbilicus and 3 cm medial to the right iliac crest (Fig. 6). The three fascial layers are incised (Fig. 7). The lateral anastomotic branches of the deep circumflex and musculophrenic arteries and associated veins must be avoided. Exposure of the extraperitoneal space is achieved by rolling the peritoneum medially and cephalad (Fig. 8). The para-aortic area is exposed by blunt dissection and node sampling is done, removing the precaval fat pad over the vena cava. With a Deaver retractor on the medial and cephalad portion of the incision, para-aortic nodes can be sampled to the level of the inferior mesenteric artery, directly anterior and lateral to the aorta.

In the left-sided approach, left para-aortic node dissection is more easily accomplished. Berman and colleagues⁷⁴ described lifting the peritoneum from the underlying vena cava to resect the precaval fat pad. They caution that gentle traction must be used when dissecting the precaval nodes to avoid injuring the inferior mesenteric vessels. The inferior vena cava lies posterior to the aorta. We have had technical difficulties in removing precaval nodes through a left-sided approach. Also, avulsion of the inferior mesenteric artery has been reported using the left-sided J-shaped incision.⁶⁴ In some patients, the lower portion of the J-shaped incision may lie in future radiation fields.

Because of the potential delay in irradiation, we have subsequently used a supraumbilical, transverse “sunrise” incision (Fig. 9).^78,79 The incision is carried laterally in a downward fashion to the level of the iliac crests. In the event of palpable, bulky disease in the lower common iliac or pelvic nodes, the incision can be carried in a caudad fashion to remove the nodes. (The advantage of pelvic node debulking in cervical cancer patients who were later irradiated has been reported by Downey and associates.⁸⁰) The fascia is then incised in a transverse manner and the rectus muscles are transected. The transversus muscle is incised, and the incision in the transversus is carried caudad while the surgeon retracts the peritoneum medially. The surgeon's hand is inserted deep into the caudad portion of the incision (Fig. 10). In thin patients, the para-aortic nodes lateral and anterior to the aorta can be removed by a right abdominal approach. However, if exposure is limited, the peritoneum must be mobilized from the left side, as previously described.

Using this technique, patients with negative nodes had a mean number of 12.2 removed from the para-aortic area.⁷⁹ All but two patients underwent external beam irradiation within 2 weeks (Fig. 11).

Surgical staging may offer a survival advantage in a few select patients, primarily because of occult distant metastases and local failures.⁸¹ Operative morbidity in patients who are good surgical candidates is low, but intraoperative vascular injuries, use of blood transfusions, and hematoma formation requiring re-exploration has been reported.^38,82 A recent report noted the rare recurrence of skin metastases near a drain site of a patient who had surgical staging for stage IIB cervical cancer.⁸³

LAPAROSCOPIC APPROACH.

Experience with laparoscopic removal of nodes is accumulating. In one of the early reports from this country, laparoscopy was followed by laparotomy in patients with cervical cancer scheduled for radical hysterectomy and node dissection, and in patients weighing less than 200 lb, 98% of the nodes had been removed through the laparoscope.⁸⁴ The mean yield for para-aortic nodes removed bilaterally was 6.3 nodes. A subsequent report by Spirtos and colleagues⁸⁵ revealed a mean bilateral para-aortic node count of 7.9. In this series of 40 patients, 2 patients had to be explored to control bleeding. Recio and associates⁸⁶ reported the use of laparoscopic staging for large (more than 5 cm) stage IB2 cervical cancers: a mean number of 7 para-aortic nodes were removed, all patients were discharged from the hospital within 24 hours, and all began irradiation within 1 week of surgery.

Para-aortic lymph node resection through the laparoscope for cervical cancer staging has been reported from other countries, with a similar mean number of nodes removed.^87,88 In a report from Taiwan, 43% of patients who had macroscopic invasion of para-aortic nodes did not have suspicious nodes on CT scanning.⁸⁸

One major concern is that laparoscopic para-aortic node sampling is a transperitoneal technique. Although animal studies suggest that adhesions occurring in pigs undergoing extraperitoneal lymphadenectomy are similar to those undergoing transperitoneal laparoscopy, none of these animals were irradiated.⁸⁹ Another major concern is the occurrence of vena caval injuries requiring laparotomy, even in the hands of experienced laparoscopists.⁹⁰ The pressure needed for adequate visualization can result in gas embolism in the presence of a large vena caval laceration. Further, scattered reports suggest that laparoscopy can disseminate otherwise isolated pelvic disease in cervical cancer patients. A report from the University of Alabama noted squamous cell carcinoma in an umbilical trocar site.⁹¹ Pastner and Damien⁹² reported a similar occurrence, and Cohn and colleagues⁹³ reported a patient with a stage IIB cervical adenocarcinoma who developed intraperitoneal spread after laparoscopic lymphadenectomy.

Clearly, para-aortic lymphadenectomies can be safely done by expert laparoscopists in scattered areas of this country. The GOG is conducting a study of open removal of pelvic and para-aortic nodes plus abdominal hysterectomy versus laparoscopic node sampling plus laparoscopic-assisted vaginal hysterectomy. This LAP-2 study has slow accrual, perhaps because many gynecologic oncologists are not qualified to remove para-aortic nodes through the laparoscope. Investigators reporting the use of this procedure emphasize that it is an evolving technique with a learning curve.^{84,85,86,87,88,90}

SCALENE NODE SAMPLING.

The issue of sampling left supraclavicular nodes has been raised, particularly in patients with positive para-aortic nodes. In one study of 72 patients with disease of various stages, no patient had metastatic disease in the removed scalene fat pad.⁹⁴ Ketcham and associates⁹⁵ found that only 6 of 108 patients had scalene node metastases when the nodes were nonpalpable. Even in patients with positive para-aortic nodes, others have reported a similarly low rate of metastases.^96,97 Before scalene node biopsy is done, a chest CT should be performed to rule out metastases in patients with nonpalpable nodes. If the para-aortic nodes are positive, scalene node biopsy could give information useful for subsequent treatment planning.

In this cost-conscious era, and conforming to the 1995 FIGO staging, the following guidelines are suggested for studies on patients with cervical cancer, according to stage:

1. IA: Complete blood count with differential, blood urea nitrogen, creatinine, electrolytes, chest x-ray
   
2. IB1: IA studies plus IVP
   
3. IB2: IA studies plus CT or MRI (abdominal and pelvic)
   
4. IIA: IA studies plus IVP; consider cystoscopy for anterior lesions
   
5. IIB: IA studies plus CT or MRI (abdominal and pelvic); consider cystoscopy for anterior lesions
   
6. III: IA studies plus CT or MRI (abdominal and pelvic); cystoscopy for anterior lesions, proctoscopy for posterior lesions
   
7. IV: All studies mentioned above; consider chest CT.
   

Barium enema studies are reserved for older patients and bone studies for symptomatic patients. If irregularity of the bladder is noted on CT or MRI, cystoscopy is indicated.

Surgical staging of any type should be used in the absence of positive radiographic studies in highly selected patients who are good surgical risks and who understand the possible pitfalls. Some patients with metastatic disease (particularly microscopic disease) can be salvaged by extended-field irradiation, with a 27% 5-year survival rate.⁷² For most patients, the current role for surgical staging is for patients willing to enter prospective randomized trials of therapeutic modalities.