This review assessed ultrasound, magnetic resonance imaging, computed tomography (CT) and positron emission tomography (PET) for therapeutic decision making in patients with early stage cervical cancer. The authors concluded that there was a lack of evidence but noted the potential of PET/CT. Weaknesses in the search strategy, review methods and reporting mean these conclusions cannot be considered reliable.

To assess the diagnostic value of parameters considered key to therapeutic decision making in patients with early cervical cancer when assessed using modern imaging modalities.

MEDLINE was searched for studies in English. No search dates were reported. Bibliographies of retrieved articles and reviews were screened for additional studies. Search terms were reported.

Studies that investigated imaging modalities – ultrasound, positron emission tomography (PET), computed tomography (CT), or magnetic resonance imaging (MRI) – for the assessment of at least 30 patients with early stage cervical cancer were eligible for inclusion. Where multiple articles assessed the same imaging modality, only studies with at least 50 participants were included. Included studies were required to use histopathology of the surgical specimen (hysterectomy or trachelectomy, lymph node dissection) as the reference standard. Key parameters considered were tumour diameter (cut-off values of 2cm or 4cm in the largest dimension), distance between the tumour and internal os (internal orifice of the uterus), outer third stromal cervical invasion, parametrial invasion and lymph node invasion. Studies of imaging for the assessment of vaginal involvement were excluded.

Where reported, studies included patients with stage IA to IIB disease. Studies assessed imaging for the detection of pelvic nodal metastases, evaluation of parametrial invasion or stromal invasion.

The authors did not report how many reviewers performed study selection.

The authors did not state that they assessed study quality.

Data were extracted on the study design (prospective or retrospective), target condition and its prevalence, sensitivity and specificity and positive and negative predictive value.

They authors did not state how many reviewers undertook data extraction.

Studies were summarised using a narrative synthesis grouped by target condition and imaging modality.

The number of studies and the total number of study participants included in the review were unclear.

CT cannot be used to evaluate tumour size and no studies were identified which assessed tumour diameter cut-off values of 2cm or 4cm using ultrasound, MRI or PET. No studies specifically assessed imaging for distance between tumour and internal os.

Three retrospective studies assessed MRI for evaluation of complete stromal involvement. Reported sensitivities ranged from 56% to 94% and reported specificities ranged from 62% to 91%.

Three large prospective studies assessed transrectal or transvaginal ultrasound for evaluation of parametrial involvement. Reported sensitivities ranged from 67% to 83% and reported specificities ranged from 56% to 100%. Nine studies assessed MRI for this application. Reported sensitivities ranged from 38% to 100% and reported specificities ranged from 46% to 99%.

Nine studies assessed MRI for the evaluation of lymph node involvement. Reported sensitivities ranged from 19% to 64% and reported specificities ranged from 69% to 99%. Three studies assessed CT for this application. Reported sensitivities ranged from 15% to 31% and reported specificities ranged from 85% to 93%. Six studies evaluated PET or PET/CT and reported sensitivities ranged from 10% to 91% and reported specificities ranged from 84% to 100%.

Data were lacking on the performance of imaging modalities in selecting treatment options for patients with early cervical cancer. The role of PET/CT in assessing lymph node invasion was promising but required further investigation.

The stated aim of this review was to assess the diagnostic value of various imaging parameters and modalities for therapeutic decision making in early stage cervical cancer. The performance of imaging modalities was assessed in relation to a histopathological reference standard but accuracy in selecting treatment options was not directly addressed.

The literature search was limited to one bibliographic database and studies in English so relevant studies may have been missed. The authors did not report any measures to minimise error and bias in the review process and did not provide any assessment of the methodological quality of the included studies. Therefore, the possible impact of flaws in the review process or poor quality primary studies could not be assessed.

Details of the included studies were poorly reported and it was unclear how many studies and participants were included in the review. The authors' conclusion that there was a lack of evidence on the performance of imaging modalities reflects the data presented but it is not clear from the results why PET/CT was highlighted as particularly promising. Given the limitations in the search strategy, review methods and reporting, these conclusions cannot be considered reliable.

Research: The authors stated that further prospective large studies were needed, particularly on the diagnostic value of normal sized hypermetabolic lymph nodes seen on PET/CT.

Practice: The authors did not specify any recommendations for clinical practice.

Not stated.